Processing of Anti-Müllerian Hormone Regulates Receptor Activation by a Mechanism Distinct from TGF-β

Abstract

TGF-β family ligands are translated as prepropeptide precursors and are processed into mature C-terminal dimers that signal by assembling a serine/threonine kinase receptor complex containing type I and II components. Many TGF-β ligands are secreted in a latent form that cannot bind their receptor, due to the pro-region remaining associated with the mature ligand in a noncovalent complex after proteolytic cleavage. Here we show that anti-Müllerian hormone (AMH), a TGF-β family ligand involved in reproductive development, must be cleaved to bind its type II receptor (AMHRII), but dissociation of the pro-region from the mature C-terminal dimer is not required for this initial interaction. We provide direct evidence for this interaction by showing that the noncovalent complex binds to a soluble form of AMHRII in an ELISA format and to AMHRII immobilized on Sepharose. Binding of the noncovalent complex to Sepharose-coupled AMHRII induces dissociation of the pro-region from the mature C-terminal dimer, whereas no dissociation occurs after binding to immobilized AMH antibodies. The pro-region cannot be detected after binding of the AMH noncovalent complex to AMHRII expressed on COS cells, indicating that pro-region dissociation may occur as a natural consequence of receptor engagement on cells. Moreover, the mature C-terminal dimer is more active than the noncovalent complex in stimulating Sma- and Mad-related protein activation, suggesting that pro-region dissociation contributes to the assembly of the active receptor complex. AMH thus exemplifies a new mechanism for receptor engagement in which interaction with the type II receptor promotes pro-region dissociation to generate mature ligand.

The mature domain of AMH can bind to the type II receptor, AMHRII, while still associated with its pro-region. Binding to AMHRII induces dissociation of the pro-region.

The TGF-β superfamily of growth factors regulates many aspects of cell growth and differentiation (reviewed in Ref. 1). Within this superfamily, the bone morphogenetic proteins (BMPs) and growth differentiation factors (GDFs) have important roles in embryonic patterning and morphogenesis (1), whereas the TGF-βs regulate processes involved in injury repair, cellular proliferation, adhesion, and immunity (2). Consistent with the important roles played by these factors, a number of regulatory mechanisms have evolved that control access of TGF-β family members to their receptors (3). One of these mechanisms involves posttranslational proteolytic processing. TGF-β family ligands are translated as dimeric precursor proteins comprising two polypeptide chains, each containing a large N-terminal pro-region and a much smaller C-terminal mature domain, which must undergo cleavage at dibasic or monobasic sites located between the two domains to generate the mature protein (Fig. 1). Even after proteolytic processing at these sites, the pro-regions of many TGF-β ligands remain noncovalently associated in a complex with their mature domains. The pro-regions of TGF-β (4, 5), GDF-8 (6), and BMP-2 (7) block binding of the mature ligands to their type II receptors, thus rendering the noncovalent complexes latent (7, 8, 9, 10). A variety of mechanisms have been identified for achieving dissociation of these latent complexes to release active ligands (11, 12, 13, 14, 15, 16, 17, 18).

Fig. 1.

Schematic diagram showing processing of a TGF-β family ligand. Processing is shown for a homodimeric precursor as is the case for AMH, but heterodimeric precursors also exist for other TGF-β family members.

Anti-Müllerian hormone (AMH), also called Müllerian-inhibiting substance, is a member of the TGF-β family responsible for the regression of Müllerian ducts in the male embryo (reviewed in Ref. 19). In female embryos, the Müllerian ducts give rise to the uterus, Fallopian tubes, and upper part of the vagina (20). In the adult, AMH also plays a role in Leydig cell differentiation and function (21, 22) and follicular development (23). Like TGF-β, AMH is produced as a large homodimeric precursor. Although only a small amount (5–20%) of the precursor is cleaved at monobasic sites upstream of the mature domains after secretion from bovine Sertoli cells (24) or cell lines (25, 26), proteolytic processing can be driven to completion in vitro (25, 26). Cleavage at the monobasic sites generates 110-kDa N-terminal and 25-kDa C-terminal homodimers, which remain associated in a noncovalent complex (Fig. 1) (25). In contrast to TGF-β, GDF-8, and BMP-2, the AMH noncovalent complex is biologically active (27), as are the noncovalent complexes of BMP-7 (28) and BMP-9 (29). Furthermore, the AMH C-terminal homodimer is much less active than the noncovalent complex in some biological assays, but almost full activity can be restored by adding back the N-terminal pro-region, which reforms a complex with the mature C-terminal dimer (27). This finding raises the possibility that the AMH noncovalent complex is the active form of the protein.

Like other members of the TGF-β family, AMH signals by assembling a transmembrane serine/threonine kinase receptor complex of type I and type II components, resulting in the phosphorylation and activation of type I receptor kinase by the constitutively active kinase domain of the type II receptor (30). The activated type I receptor then phosphorylates the cytoplasmic Sma- and Mad-related proteins (Smads) 1, 5, or 8, which migrate into the nucleus and, in concert with other transcription factors, regulate responsive genes (reviewed in Ref. 31). The type II receptor for AMH (AMHRII) is one of five type II receptors in the TGF-β family. AMH and AMHRII are mutually specific (32, 33). In contrast, the other four type II receptors for activin (ActRIIA, ActRIIB), BMP (BMPRII), and TGF-β (TβRII), each interact with multiple ligands, and all other TGF-β family ligands interact with multiple type II receptors. Two BMP type I receptors, activin receptor-like kinase (ALK)-2 (34, 35) and ALK-3 (36), can serve as type I receptors for AMH (37). ALK-6 also binds AMH (38) but acts as a negative regulator of intracellular signaling (39).

In this report, we show that the cleaved AMH noncovalent complex binds to AMHRII and stimulates intracellular signaling, whereas full-length AMH shows only minimal activity. Most significantly, our analyses with AMHRII immobilized on a surface or expressed on COS cells indicate that binding of the noncovalent AMH complex to the receptor causes the pro-region to dissociate. AMH thus exemplifies a new mechanism for receptor engagement in which interaction with the receptor itself promotes pro-region dissociation to generate mature ligand.

Results

Cleaved and C-terminal AMH stimulate Smad phosphorylation

Full-length, cleaved, and C-terminal AMH (Fig. 1) were prepared as previously described (25, 27). The full-length AMH preparation used in the current study comprises approximately 85% uncleaved 140-kDa homodimer plus about 15% cleaved, noncovalent complex of N- and C-terminal homodimers arising from processing by endogenous proteases during expression. Fully cleaved AMH was generated from the full-length protein by digestion with plasmin and contains 100% noncovalent complex. C-terminal AMH (C-terminal homodimer) was prepared by separation of plasmin-cleaved AMH by gel filtration chromatography (27). A mouse Sertoli cell line, SMAT1 (40), which expresses AMHRII and all three type I receptors (ALK-2, ALK-3, and ALK-6) and is responsive to AMH as evident by phosphorylation of Smad1, -5, and -8, was used to assess which forms of AMH can stimulate intracellular signaling (39).

Full-length, cleaved, and C-terminal AMH were incubated with SMAT1 cells at various concentrations and for varying times, and their ability to induce Smad phosphorylation was assessed by Western blotting of cell lysates. Figure 2, A and C, show a dose response and time course for Smad phosphorylation induced by C-terminal, cleaved, and full-length AMH. In the dose-response study, where the AMH treatment was for only 30 min, C-terminal and cleaved AMH showed significant dose-dependent stimulation of Smad phosphorylation, whereas full-length AMH showed much less activity (Fig. 2A). A densitometry analysis of the autoradiograph in Figure 2A is shown in Figure 2B and confirms that both C-terminal and cleaved AMH are more potent than full-length AMH in stimulating Smad phosphorylation. Furthermore, Figure 2B shows that C-terminal AMH is 3- to 5-fold more active than cleaved AMH when the two preparations are compared on a mole-for-mole basis.

Fig. 2.

C-terminal and cleaved AMH can stimulate Smad phosphorylation but full-length AMH has minimal activity. A, Dose-response comparison of Smad phosphorylation induced by C-terminal, cleaved, and full-length AMH. SMAT1 cells were treated for 30 min with the indicated form of AMH at various concentrations and phospho-Smad1/5/8, total Smad1/5/8, and tubulin levels were assessed by Western blotting of cell lysates. B, The Smad1/5/8-P band intensities in panel A above basal level were determined by densitometry. C, Time course comparison of Smad phosphorylation induced by C-terminal, cleaved, and full-length AMH. SMAT1 cells were treated with the indicated form of AMH (8 nm) for various times, and phospho-Smad1/5/8, total Smad1/5/8, and tubulin levels were assessed by Western blotting of cell lysates. D, The Smad1/5/8-P band intensities in panel C above basal level were determined by densitometry. C-terminal AMH induced a higher level and a faster rate of Smad phosphorylation than cleaved AMH, whereas full length AMH showed only minimal activity, consistent with it being due to contaminating cleaved AMH present in the full-length AMH preparation. Results shown are representative of at least three independent experiments.

In the time course study, in which the AMH concentration was 8 nm, C-terminal and cleaved AMH again showed more activity than full-length AMH in stimulating Smad phosphorylation (Fig. 2, C and D). The levels of Smad phosphorylation induced by the full-length AMH preparation were less than 20% of the signals produced by cleaved AMH at each time point, and is likely due to the approximately 15% of contaminating cleaved AMH that is present in the full-length protein preparation. Thus, these experiments show that cleaved AMH is active, but that uncleaved AMH is unable to stimulate Smad phosphorylation in SMAT1 cells. Similar to what was observed in the dose-response study, C-terminal AMH was more active than cleaved AMH and induced Smad phosphorylation at a faster rate than was produced by an equimolar concentration of cleaved AMH. These results show that the N-terminal pro-region of AMH is not required for activity on these cells, and indeed appears to attenuate the activity of the C-terminal homodimer.

Soluble AMHRII (AMHRII-Fc) binds to cleaved AMH but not to full-length AMH

The results of the previous experiments indicate that cleaved and C-terminal AMH can induce Smad phosphorylation. Because the ability of a ligand to activate an intracellular pathway depends on the assembly of an activated receptor complex on the cell, we investigated the binding of AMH and its cleavage products to AMHRII. AMHRII-Fc was produced as shown in Figure 3. The fusion protein, containing the extracellular domain of human AMHRII fused to the Fc portion of human IgG1 (Fig. 3A), was expressed in 293E cells and purified using Protein A Sepharose. The gel analysis shown in Figure 3B indicated that the fusion protein consists mostly of dimer and a smaller amount of tetramer.

Fig. 3.

Generation of an AMHRII-Fc fusion protein. A, Schematic diagram showing AMHRII and AMHRII-Fc fusion proteins. The numbering corresponds to amino acid residues in UniProtKB accession numbers Q16671 (human AMHRII) and P01857 (human Ig γ-1 chain C region). ECD, Extracellular domain; TM, transmembrane domain. B, SDS-PAGE analysis of the AMHRII-Fc fusion protein analyzed under reducing and nonreducing conditions and detected by staining with Coomassie blue. The AMHRII-Fc fusion protein consists mostly of dimer and a smaller amount of tetramer.

Binding of AMH to AMHRII-Fc was assessed using the ELISA format shown in the schematic diagram in Figure 4A. Assay plates were coated with the mouse monoclonal antibody (mAb) 10.6, which recognizes an epitope on the pro-region of human AMH. Four different forms of AMH could then be captured by this monoclonal antibody: full length, cleaved, and noncleavable AMH and pro-region. The pro-region preparation contains only the N-terminal homodimer, whereas the biologically inactive noncleavable AMH preparation contains uncleaved 140 kDa homodimer due to R451T mutations that render the monobasic cleavage sites resistant to proteolysis (27, 41). Assay wells containing each of these different forms of AMH captured via the antibody were incubated with the AMHRII-Fc fusion protein, and then bound AMHRII-Fc was detected with an antihuman Fc secondary antibody. As shown in Figure 4A, only cleaved AMH showed significant binding to the AMHRII-Fc protein. The full-length AMH preparation reproducibly showed a low level of binding to AMHRII-Fc that was always higher than that produced by noncleavable AMH or pro-region, which is consistent with the presence of approximately 15% cleaved protein in the full-length AMH preparation.

Fig. 4.

AMHRII-Fc binds to cleaved AMH but not to full-length AMH. A, The schematic shows the ELISA setup used to assess binding of AMHRII-Fc to cleaved, noncleavable, and full-length AMH and pro-region. AMH at various concentrations was captured by anti-pro-region mAb 10.6 coated on the ELISA plate, incubated with AMHRII-Fc (64 nm), and bound AMHRII-Fc was detected with a goat antihuman Fc antibody conjugated to HRP. In the ELISA results on the right, only cleaved AMH showed significant binding to AMHRII-Fc. B, The schematic shows the ELISA setup used to verify that mAb 10.6 captured equivalent amounts of the four AMH preparations. An anti-AMH polyclonal antibody, which recognizes mostly epitopes on the pro-region, was used in place of AMHRII-Fc. Results shown are representative of at least three independent experiments. Absorbance values are reported as millioptical density units.

To verify that all four AMH proteins were present in the assay at equivalent levels, parallel mAb 10.6 coated plates were treated with the four AMH proteins and analyzed directly for the presence of bound AMH using an anti-AMH polyclonal antibody that recognizes primarily the pro-region domain (schematic diagram; Fig. 4B). The ELISA results (Fig. 4B) show that similar amounts of all four AMH preparations were captured by the anti-pro-region mAb 10.6. Thus, the low signal observed for the full-length AMH preparation in Figure 4A and the absence of signal for noncleavable AMH and pro-region is not due to a failure of these proteins to be captured by the anti-AMH monoclonal antibody, but instead indicates an inability to bind strongly to AMHRII.

The assays shown in Figure 4A were performed by varying the AMH concentration and holding the AMHRII-Fc concentration constant. To investigate whether the interaction of AMHRII-Fc with cleaved AMH was saturable, the format was switched so that the AMH concentration was held constant and the AMHRII-Fc concentration was varied. As shown in Figure 5A, significant binding of cleaved AMH to AMHRII-Fc that was saturable was observed, whereas a much lower signal was observed for the full-length AMH preparation. At high AMHRII-Fc concentrations the signal for full-length AMH was higher than that produced by noncleavable AMH consistent with it being produced by contaminating cleaved AMH. To demonstrate the specificity of binding of cleaved AMH to AMHRII-Fc, an experiment was performed using other type II receptor-Fc fusion proteins in place of AMHRII-Fc. As shown in Figure 5B, only AMHRII-Fc bound to cleaved AMH; no binding was observed with the soluble type II receptors for TGF-β, BMP, and activin (TβRII-Fc, BMPRII-Fc, and ActRIIB-Fc). Taken together, the ELISA experiments show that cleaved AMH can specifically bind to AMHRII-Fc, whereas uncleaved or noncleavable AMH or pro-region alone cannot.

Fig. 5.

Binding of AMHRII-Fc to cleaved AMH is saturable and specific. A, Cleaved, noncleavable, and full-length AMH (16 nm) were captured by anti-pro-region mAb 10.6, incubated with AMHRII-Fc at various concentrations, and bound AMHRII-Fc was detected with a second antibody as shown in the schematic on the left. In the ELISA results on the right, AMHRII-Fc showed saturable binding to cleaved AMH but lesser or no binding to the other AMH preparations. B, Cleaved AMH (40 nm) was captured by mAb 10.6, incubated with AMHRII-Fc, TβRII-Fc, BMPRII-Fc, or ActRIIB-Fc at various concentrations, and bound receptor-Fc was detected with a second antibody as shown in the schematic on the left. Only AMHRII-Fc showed binding to cleaved AMH. Results shown are representative of at least three independent experiments.

AMHRII-Fc binds to cleaved and C-terminal AMH with similar affinities

To investigate whether AMHRII-Fc also binds to the isolated C-terminal homodimer of AMH, the AMHRII binding ELISA format was changed to use an antibody that recognizes the C-terminal domain of AMH (mAb 22A2) to capture AMH (schematic diagram; Fig. 6A). As shown in Figure 6A, binding of AMHRII to both C-terminal and cleaved AMH was observed, whereas little binding was observed to full-length or noncleavable AMH. Additionally, this result demonstrates that mAb 22A2 does not interfere with AMHRII binding. On a parallel mAb 22A2-coated plate, the AMH proteins were directly detected with an anti-AMH rabbit polyclonal antibody to assess whether they were present in the AMHRII-binding ELISA at equivalent levels (schematic diagram, Fig. 6B). Figure 6B shows that similar amounts of cleaved, full-length, and noncleavable AMH were captured by the anti-C-terminal AMH antibody. A lower signal for C-terminal AMH occurs because the anti-AMH polyclonal antibody is less reactive toward the C-terminal domain than the pro-region domain (data not shown). This control establishes that the absence of signal seen in Figure 6A for full-length or noncleavable AMH is not due to a failure to capture these proteins on the assay plate.

Fig. 6.

AMHRII-Fc binds to C-terminal AMH. The schematic shows the ELISA setup used to assess binding of AMHRII-Fc to cleaved, noncleavable, full-length, and C-terminal AMH. AMH at various concentrations was captured by anti-C-terminal mAb 22A2 coated on the ELISA plate, incubated with AMHRII-Fc (64 nm), and bound AMHRII-Fc was detected with a goat antihuman Fc antibody conjugated to HRP. In the ELISA results on the right, C-terminal and cleaved AMH showed significant binding to AMHRII-Fc. B, The schematic shows the ELISA setup used to verify that mAb 22A2 captured equivalent amounts of cleaved, noncleavable, and full-length AMH. Results with C-terminal AMH are not shown because it is not detected very well by the anti-AMH polyclonal antibody. Results shown are representative of at least three independent experiments.

In Figure 7A, the ELISA format was switched from that used in Figure 6A such that the AMH concentration was held constant and the AMHRII-Fc concentration was varied. In this experiment, AMHRII-Fc again bound to both cleaved and C-terminal AMH, and a low level of binding of AMHRII-Fc was again observed to the full-length AMH preparation. As was observed in Figure 6A, C-terminal AMH gave a somewhat higher signal at saturation than cleaved AMH; however, the concentrations of AMHRII-Fc at which half-maximal binding to the two proteins occurred were both around 10 nm, indicating that the affinities for the interactions are probably similar. To demonstrate that the binding of C-terminal AMH to AMHRII-Fc is specific, another experiment was performed using other type II receptor-Fc fusion proteins in place of AMHRII-Fc. As shown in Figure 7B, only AMHRII-Fc was able to bind to C-terminal AMH; no binding was observed with the soluble type II receptors for TGF-β, BMP, or activin.

Fig. 7.

Binding of AMHRII-Fc to C-terminal AMH is saturable and specific. A, C-terminal, cleaved, noncleavable, and full-length AMH (40 nm) were captured by anti-C-terminal mAb 22A2, incubated with AMHRII-Fc at various concentrations, and bound AMHRII-Fc was detected with a second antibody as shown in the schematic on the left. In the ELISA results on the right, AMHRII-Fc showed saturable binding to C-terminal and cleaved AMH with half-maximal binding occurring at around 10 nm for both proteins. The absorbance values obtained for noncleavable AMH were used as baseline because of higher nonspecific background to the non-AMH-treated plates at high AMHRII-Fc concentrations. B, C-terminal AMH (40 nm) was captured by mAb 22A2, incubated with AMHRII-Fc, TβRII-Fc, BMPRII-Fc, or ActRIIB-Fc at various concentrations, and bound receptor-Fc was detected with a second antibody as shown in the schematic on the left. Only AMHRII-Fc showed binding to C-terminal AMH. Results shown are representative of at least three independent experiments.

Selective dissociation of the pro-region from the C-terminal homodimer after binding of cleaved AMH to AMHRII

The interaction between AMHRII and cleaved AMH was further investigated in a precipitation experiment using AMHRII-Fc coupled to Sepharose. After incubating full-length, cleaved, and noncleavable AMH and AMH pro-region with the receptor-coupled Sepharose beads, the resin was collected and washed, and bound protein was eluted and detected by Western blotting using a polyclonal antibody that recognizes the pro-region domain. This antibody was used to detect precipitated proteins because the pro-region is present in all four of the species being tested. Figure 8A shows that only cleaved AMH efficiently bound to the AMHRII-Fc-coupled Sepharose, as evidenced by detection of the pro-region. In contrast, very little of the preparation containing only AMH pro-region was precipitated, as expected given that AMHRII doesn’t bind the pro-region of AMH. Also, relatively little full-length or noncleavable AMH was precipitated. Thus, only the cleaved AMH could interact with the AMHRII on the beads, an interaction that the results in Figs. 4–7 show is mediated through the C-terminal domain of this complex. Moreover, the binding of AMH to AMHRII cannot occur until the full-length protein has undergone proteolytic cleavage.

Fig. 8.

Binding of AMHRII-Fc to cleaved AMH promotes dissociation of the pro-region. A, Cleaved, full-length, and noncleavable AMH and pro-region were precipitated by AMHRII-Fc-Sepharose, washed once, and analyzed by Western blotting under nonreducing conditions with an anti-AMH polyclonal antibody that recognizes the pro-region domain. Only cleaved AMH was efficiently precipitated by the AMHRII-Fc-Sepharose as indicated by the recovery of the pro-region dimer. B, Cleaved AMH was precipitated by AMHRII-Fc-Sepharose, washed one to three times, and analyzed by SDS-PAGE under nonreducing conditions. The pro-region dimer was detected by Western blotting with an anti-AMH polyclonal antibody, whereas C-terminal dimer was detected by protein staining. In the top panel, an equivalent amount of full-length AMH was run along with the cleaved AMH in the lane containing AMH before precipitation. Multiple washes resulted in loss of pro-region but not C-terminal dimer after precipitation of cleaved AMH. C, Biotinylated-cleaved AMH was precipitated with the indicated reagent, washed two times, and analyzed by Western blotting under reducing conditions using a streptavidin-HRP conjugate. The image was captured using a Kodak Image Station. A higher amount of protein was loaded in lane 2 compared with the other lanes to ensure that both precipitated species could be clearly seen. 2° Ab, secondary antimouse antibody coupled to magnetic beads. D, Band intensities for the pro-region and C-terminal monomers shown in panel C were measured using Kodak Molecular Imaging Software, and the ratio of C-terminal and pro-region monomers was calculated for each sample. The ratio of C-terminal and pro-region monomers before precipitation has been set to 1, and the ratios of the precipitated samples have been normalized by this factor. The ratio for the C-terminal and pro-region monomers is close to 1 after precipitation with mAbs 10.6 or 22A2, whereas it is above 8 after precipitation with AMHRII-Fc, indicating that the AMH complex stays intact during precipitation with antibodies to either the pro-region or C-terminal dimer, but undergoes significant dissociation of the pro-region during precipitation with AMHRII-Fc.

The ability of cleaved AMH to bind to AMHRII-Fc was further investigated by assessing the recovery of both the pro-region and the C-terminal dimer. In this experiment, precipitated C-terminal dimer was visualized directly by staining, whereas the pro-region was detected by Western blot due to its lower staining intensity. Figure 8B shows that, as expected, both the N-terminal pro-region and the C-terminal dimer were detected after precipitation of cleaved AMH with AMHRII-Fc. However, Figure 8B also shows that as the number of washing steps was increased, the amount of precipitated pro-region decreased, whereas the level of precipitated C-terminal dimer remained constant. These results suggest that binding of AMHRII to cleaved AMH can lead to dissociation of the pro-region from the C-terminal homodimer.

To demonstrate that dissociation of the pro-region is induced specifically by the binding of AMHRII and not by the binding of other proteins such as AMH antibodies, cleaved AMH was biotinylated and subjected to precipitation by AMHRII-Fc, anti-pro-region mAb 10.6, and anti-C-terminal mAb 22A2. The pro-region and C-terminal fragment were then detected simultaneously by Western blotting using a streptavidin-horseradish peroxidase (HRP) conjugate. As shown in Figure 8C, precipitation with antibodies to either the N-terminal pro-region or C-terminal dimer did not affect the ratio of the pro-region and C-terminal monomers detected by Western blotting under reducing conditions, indicating that no dissociation had occurred during precipitation. This was true for precipitation with the anti-pro-region mAb 10.6 regardless of whether it was conjugated to Sepharose or combined with a second antibody coupled to magnetic beads (2° Ab). In contrast, precipitation with AMHRII-Fc Sepharose resulted in a significant enrichment for the C-terminal dimer. Note that a higher amount of protein was loaded in lane 2 compared with the other lanes to ensure that both precipitated species could be clearly seen. The images in Figure 8C were captured by a Kodak Image Station allowing a precise measurement of the band intensities for the pro-region and C-terminal monomers. This quantitation is shown in Figure 8D, where the ratio of C-terminal and pro-region monomers before precipitation (shown in Fig. 8C, lane 1) has been set to 1, and the ratios for the precipitated samples have been normalized by this factor. The C-terminal/pro-region ratio is close to 1 after precipitation with mAbs 10.6 and 22A2, whereas it is above 8 after precipitation with AMHRII-Fc. Thus, the AMH noncovalent complex remains intact during precipitation with an antibody to the C-terminal dimer but undergoes significant dissociation of the pro-region during precipitation with AMHRII-Fc, which also interacts with the C-terminal dimer.

Further evidence that the noncovalent complex remains intact after binding to mAb 22A2 is provided by the AMH ELISA shown in Figure 6B, where the anti-AMH polyclonal antibody detects primarily the pro-region. In this experiment, the signal observed with cleaved AMH was comparable to the signals seen with full-length and noncleavable AMH, where the fragments are covalently linked and the pro-region cannot undergo dissociation. Thus, even though the washing procedure in the ELISA protocol is significantly more stringent than that used for the precipitation experiments, no dissociation of the pro-region was observed after cleaved AMH was captured by the anti-C-terminal mAb 22A2.

The pro-region cannot be detected after binding of cleaved AMH to AMHRII on COS cells

To investigate the interaction of cleaved AMH with AMHRII on cells, binding experiments were performed on COS cells transfected with AMHRII. In these experiments, bound AMH was detected with either anti-C-terminal mAb 22A2 or anti-pro-region mAb 10.6 in combination with a fluorescein isothiocyanate (FITC)-conjugated antimouse antibody. If cleaved AMH remains intact upon binding to the AMHRII on cells, then it should be detectable using either of these two antibodies, because the ELISA results showed that neither antibody binds competitively with respect to AMHRII-Fc (Figs. 4A and 6A). When transfected cells were incubated with cleaved AMH, a strong binding signal (green) was observed using anti-C- terminal mAb 22A2 for detection (Fig. 9A). As expected, only a faint signal was observed with the full-length AMH preparation (Fig. 9C), and no binding of noncleavable AMH was observed (Fig. 9E). These results show that cleaved AMH can bind specifically to AMHRII on the transfected COS cells; the full-length and noncleavable proteins serve as controls to rule out the possibility that the binding observed with cleaved AMH is nonspecific. As an additional control for nonspecific binding we showed that no binding of cleaved AMH was observed to untransfected COS cells (data not shown).

Fig. 9.

The pro-region cannot be detected after binding of cleaved AMH to AMHRII on COS cells. A–F, COS cells were transfected with AMHRII and incubated with either cleaved AMH (A and B), full-length AMH (C and D), or noncleavable AMH (E and F). AMH bound to cells was detected with either anti-C-terminal mAb 22A2 (A, C, and E) or anti-pro-region mAb 10.6 (B, D, and F) in combination with a FITC-conjugated antimouse IgG antibody (green). G and H, COS cells were transfected with AMH, and intracellular AMH was detected with either mAb 22A2 (G) or mAb 10.6 (H). Nuclei were stained with propidium iodide (red). Cleaved AMH bound to AMHRII on COS cells could be detected with mAb 22A2 but not mAb 10.6, whereas AMH expressed in COS cells could be detected with both antibodies. Results shown are representative of at least three independent experiments.

In contrast to the results obtained with anti-C-terminal mAb 22A2, no significant signal was observed when anti-pro-region mAb 10.6 was used for detection after binding of cleaved AMH to the AMHRII transfected cells (Fig. 9B). Multiple experiments with cleaved AMH have always yielded the same result: a good signal is obtained when mAb 22A2 is used for detection, whereas no signal is obtained using mAb 10.6. To demonstrate that both of these antibodies work equivalently well in this protocol, we used a similar method to show that both mAbs can detect AMH when it is directly transfected into COS cells (Fig. 9, panels G and H). Moreover, both mAbs bind cleaved AMH comparably well in an ELISA format, whether the protein is coated directly on the assay plate or captured via a polyclonal antibody (data not shown). These results indicate that the pro-region is not present in the complex formed when cleaved AMH binds to AMHRII on cells and thus are consistent with the notion that binding of the cleaved AMH complex to AMHRII on cells promotes dissociation of the pro-region as was observed using AMHRII-Fc in vitro.

Discussion

We have used a soluble type II receptor and an AMH-responsive cell line to identify the forms of AMH that can bind to AMHRII and stimulate phosphorylation of Smads 1, 5, and 8, which represents the first step in intracellular signaling. The results showed that cleaved and C-terminal AMH can bind to AMHRII and stimulate phosphorylation of Smads, whereas noncleavable full-length AMH and also the N-terminal pro-region alone were completely inactive. Full-length AMH had low activity, which presumably results from contaminating cleaved AMH in the preparation because the level of activity correlates with the level of cleaved AMH. The ability of AMHRII to bind cleaved AMH but not full-length AMH indicates that AMH must be cleaved for it to interact with its type II receptor. AMHRII-Fc was observed to bind to cleaved AMH whether the latter was captured onto the assay plate via the pro-region or the C-terminal dimer. Moreover, the pro-region could be coprecipitated with the C-terminal homodimer using AMHRII-Fc coupled to Sepharose. These results prove conclusively that cleaved AMH can bind to AMHRII as an intact complex and rules out the possibility that the N-terminal pro-region must first dissociate from the mature ligand for the latter to bind to the type II receptor. Furthermore, the binding affinities of cleaved and C-terminal AMH for AMHRII-Fc appeared to be similar in the ELISA format (Fig. 7A), indicating that the pro-region does not significantly attenuate binding of the C-terminal dimer to AMHRII.

When cleaved AMH was added to immobilized AMHRII-Fc, we discovered that binding led to dissociation of the pro-region from the C-terminal domain and preferential loss of the pro-region from the complex. This observation is specific to interaction with AMHRII. Precipitation experiments with biotinylated-cleaved AMH showed that significant loss of pro-region was observed after precipitation with AMHRII-Fc, but no dissociation of the noncovalent complex was observed after precipitation with either pro-region- or C-terminal-specific antibodies. AMHRII on cells also appears to mediate dissociation of the N-terminal pro-region, because the pro-region was no longer detectable after binding of cleaved AMH to AMHRII expressed on COS cells. Together, these observations suggest that binding of the noncovalent cleaved AMH complex to AMHRII, when the receptor is captured on a solid surface or presented on cells, promotes dissociation of the pro-region from the C-terminal dimer. Interestingly, there was no evidence for dissociation when cleaved AMH was bound to AMHRII-Fc in ELISA experiments (Figs. 4A and 5A). A difference between these experiments is that in the two formats in which AMHRII-dependent dissociation of the cleaved AMH was observed, the receptor is presented on a surface, whereas in the ELISA experiments, it participates as a soluble protein, where it is not capable of engaging in the multivalent interactions available to a surface-bound receptor (42). Such multivalent interactions may be critical for pro- region dissociation. In theory, the AMRHII-Fc dimer is capable of binding both monomers in one C-terminal dimer, but our present working model is that this does not occur, probably due to constraints imposed by the Fc moiety. Importantly, our observation that C-terminal AMH is more active than cleaved AMH in stimulating Smad phosphorylation indicates that the pro-region attenuates the activity of the C-terminal dimer, consistent with the idea that dissociation of the pro-region contributes to the assembly of a fully active receptor complex.

The pro-regions of TGF-β (4), GDF-8 (6), and BMP-2 (7) have been shown to completely prevent binding of the mature ligand to the type II receptor. However, it is not clear whether this interference is due to direct steric interference between pro-region and receptor, or alternatively whether the bound pro-region acts allosterically by maintaining the mature C terminus in a conformation that is unable to bind receptor, because both mechanisms would appear as competitive inhibition in binding experiments. Circular dichroism measurements with TGF-β suggest an allosteric effect where the pro-region induces a conformational change after binding to mature TGF-β (43). This hypothesis has been invoked to explain how the TGF-β pro-region prevents binding of both type II and III receptors, which interact at distinct sites on the mature ligand (4). Our results are also consistent with an allosteric interaction between the receptor and pro-region binding sites on mature TGF-β family ligands, because binding of AMHRII to cleaved AMH promotes dissociation of the pro-region. Thus, despite the very different ways in which the pro-regions of TGF-β and AMH affect type II receptor binding, they may operate via a common mechanism. Their differential effects on receptor binding could be a consequence of the nature of the conformational change induced in the mature ligand by the pro-region, with it being sufficient in the case of TGF-β to completely prevent binding to the type II receptor, but in the case of AMH leading only to a minimal reduction in type II receptor binding.

Figure 10 shows a model for the assembly of the AMH receptor signaling complex that is consistent with our observations. In full-length AMH, the C-terminal domain is presented in a conformation that is unable to bind receptor. Proteolytic cleavage allows the C-terminal domain to adopt a conformation that can bind to AMHRII. Binding of AMHRII to cleaved AMH allosterically weakens binding of the pro-region to the C-terminal domain, promoting its dissociation. A positive feature of this allosteric model is that it can potentially account for the differences in the biochemical behavior of AMH and TGF-β without invoking fundamentally distinct binding modes between the respective pro-regions and mature ligands; in contrast, distinct binding modes would be required if the AMH pro-region were to operate via a steric interference mechanism whereas the TGF-β pro-region operates via an allosteric mechanism. Sengle et al. (28) recently proposed a steric interference model for the affect of the BMP-7 pro-region on type II receptor engagement, inferred from the sedimentation properties of the BMP-7 noncovalent complex when combined with BMPRII-Fc in sucrose gradients. To explain the formation of a ternary complex between mature ligand, pro-region, and type II receptor, they proposed the establishment of a complex in which one pro-region monomer has dissociated from one C-terminal monomer, thus making room for one type II receptor molecule to bind. However, the existence of a stable 2:1 complex between a mature C-terminal dimer and a single pro-region monomer has not been shown for any TGF-β family ligand. The AMH pro-region is a disulfide-linked dimer, and therefore it is highly unlikely that a single pro-region monomer would dissociate from one C-terminal monomer without the N-terminal pro-region undergoing full dissociation. We therefore believe that the allosteric mechanism that we have proposed provides a more likely interpretation of our results for AMH and indeed provides an alternative explanation for the results reported by Sengle et al. (28).

Fig. 10.

Model showing how processing of AMH may regulate the assembly of the receptor signaling complex. Cleavage of full-length AMH results in a conformational change in the C-terminal domain, indicated by the shape and color change, which allows binding of AMRHII. Binding of AMHRII induces dissociation of the pro-region via a negative allosteric interaction between the receptor- and pro-region-binding sites on the C-terminal dimer, indicated by the shape change. Results presented in this paper are consistent with pro-region dissociation occurring before type I receptor engagement, but this has not been proven. The type I and II receptor-binding sites on the C-terminal dimer are indicated by either a I or a II; black labels indicate sites on the front of the dimer, and white labels indicate sites on the back of the dimer.

How AMH interacts with the type I receptor remains to be elucidated. Regression of the Müllerian duct has been shown to require ALK-3 (36, 37), but so far we have been unable to demonstrate an interaction between ALK-3 and any form of AMH (di Clemente, N., and R.Cate, unpublished results). Although it has been reported that the BMP-7 noncovalent complex can bind ALK-3 in solution (28), mature BMP-7 shows little binding to ALK-3 on cells in the absence of a type II receptor (44, 45) and requires a cooperative interaction with ActRII to bind ALK-2 (46). In surface plasmon resonance studies, mature BMP-7 binds to immobilized ALK-3 with an apparent dissociation constant (K_D) of 58 nm (47). Our inability to detect an interaction between AMH and ALK-3 indicates that the binding of AMH to ALK-3 is weak. Also, dissociation of the pro-region after binding of the cleaved AMH to AMHRII on COS cells appears to be independent of the type I receptor, because transfected AMHRII should be present in large excess compared with any endogenous type I receptor on these cells. Therefore, for AMH, it is likely that the type I receptor is recruited into the signaling complex after cleaved AMH has bound to AMHRII and the pro-region has dissociated (Fig. 10). The lower activity of cleaved AMH compared the C-terminal dimer in the Smad signaling assay (Fig. 2) could be due to the pro-region interfering with type I receptor engagement, but at present we do not have any definitive support for this hypothesis.

Previous observations in the literature have led to some confusion concerning what form of AMH interacts with its receptors to trigger intracellular signaling. For example, full-length AMH has almost an identical ED₅₀ to that seen with cleaved AMH in assays measuring repression of aromatase production in rat fetal ovary (27, 48). However, this observation was made in a 3-d organ culture assay in which AMH proteolysis can occur. Another observation that has caused confusion is that the purified C-terminal dimer is 10-fold less active than cleaved AMH in the fetal ovary assay, but almost full activity can be restored by adding back the pro-region, which reforms a complex with the C-terminal dimer (27). This result is contrary to our observation in the Smad-signaling assay, which lasts for only 30–90 min, that the C-terminal dimer alone was more active than cleaved AMH. A likely explanation for this apparent discrepancy is that the isolated C-terminal dimer is unstable in the 3-d assay, and the pro-region helps stabilize it until it binds to AMHRII. Indeed, the biochemical properties of the AMH C-terminal dimer after dissociation from the pro-region make it unlikely that it exists in a free form, because it tends to aggregate at neutral pH and binds nonspecifically to various surfaces (27). Similar observations have been made with other TGF-β family members, in which the pro-region has been shown to either stabilize the mature ligand (49) or increase its solubility (50).

The noncovalent complexes of TGF-β, GDF-8, and BMP-2 are biologically latent due to the ability of the pro-regions to block binding of the mature ligands to their type II receptors. Dissociation of these latent noncovalent complexes is carefully regulated by extrinsic biological processes, ensuring that the active mature ligand is only released at a location where it is required. In contrast, AMH and BMP-7, and perhaps BMP-9 as well, might constitute a group of ligands within the TGF-β family that uses a novel mechanism for receptor engagement in which interaction of the noncovalent complex with the type II receptor promotes pro-region dissociation to generate mature ligand. Our findings thus suggest that the TGF-β family can be divided into mechanistic subclasses, based on distinctions in how the pro-region functions to stabilize and potentiate the mature growth factor and modulates its interaction with the type II receptor.

Materials and Methods

Generation of the AMHRII-Fc fusion protein

A cDNA encoding the signal sequence of TβRII (residues 1–23 of UniProtKB Accession no. P37173) fused to the mature extracellular domain of AMHRII (residues 18–145 of UniProtKB Accession no. Q16671) was generated by PCR, ligated to a cDNA encoding the hinge, CH2, and CH3 domains of human IgG1 (residues 104–330 of UniProtKB Accession no. P01857), and cloned into expression vector pNE001. The plasmid was transfected into 293E cells using effectene (QIAGEN, Chatsworth, CA) and medium was collected over a 2-wk period at 4-d intervals. The AMHRII-Fc fusion protein was isolated using Protein A Sepharose. BMPRII-Fc and ActRIIB-Fc fusion proteins were obtained from R&D Systems (Minneapolis, MN); human TβRII-Fc protein was the gift of Véronique Bailly (Biogen Idec).

Preparation of AMH

Full-length and noncleavable AMH and pro-region were purified by immunoaffinity chromatography from culture medium of Chinese Hamster Ovary cells transfected with the human wild-type or R451T and S452stop mutant AMH cDNAs, respectively, and stored in 10 mm HEPES (pH 7.4), 10% glucose, and 300 mm NaCl at −70 C (41). Cleaved AMH was generated by digesting full-length AMH with plasmin at a mass ratio of 40:1 for 2 h at 28 C, and the plasmin was removed using a chicken ovoinhibitor affinity column (38). C-terminal AMH was purified from cleaved AMH on a Sephacryl S-200 column equilibrated with 1 m acetic acid with 0.3 mg/ml polyethylene glycol 3350 at 4 C (27). Fractions containing the C-terminal dimer were pooled, and BSA was added at an equivalent protein concentration and stored at −70 C. Aliquots of C-terminal AMH were lyophilized and resuspended at a concentration of 100 μg/ml in 1 mm HCl, and polyethylene glycol 3350 was added to a final concentration of 3 mg/ml. Protein concentrations of AMH preparations were determined by measuring absorbance at 280 nm, using an extinction coefficient of 1.0 ml/mg cm, and confirmed by SDS-PAGE. Cleaved AMH (850 μg/ml) was biotinylated by adding NHS-LC-biotin (Thermo Fisher Scientific; Rockford, IL) to a concentration of 0.5 mm, incubating for 75 min at room temperature, and quenching by adding ethanolamine to a final concentration of 20 mm.

AMHRII binding ELISA

Dynatech Immulon 2 (96 well) ELISA plates were coated with either anti-pro-region mAb 10.6 or anti-C-terminal mAb 22A2 overnight at 4 C in 50 mm sodium bicarbonate, pH 9.6 (12.5 μg/ml; 50 μl/well). The plates were washed five times with water and then blocked for 2 h at room temperature using 150 μl/well of block buffer containing 1% BSA (A-7906; Sigma, St. Louis, MO) and 1% goat serum (16210064; Invitrogen, Carlsbad, CA) in PBS. This buffer was used for all subsequent dilutions. The block buffer was discarded and AMH was serially diluted down the plate by a factor of 2 or 3. Plates were incubated for 1 h, followed by five washes with PBS. AMHRII-Fc fusion protein (50 μl) was added to each well at a concentration of 5 μg/ml (64 nm) and incubated for 2 h. The plates were washed five times with PBS/0.05% Tween 20. Goat antihuman Fc conjugated to HRP (Jackson ImmunoResearch Laboratories, West Grove, PA) was added at a 1:5000 dilution, and the plates were incubated for 50 min. After five washes with PBS/0.05% Tween 20, 50 μl of TMB substrate were added to each well. The reactions were quenched by the addition of 50 μl/well of 2 m sulfuric acid, and absorbances were read at 450 nm. In some studies, a fixed amount of AMH (16 or 40 nm; 50 μl/well) was added to the wells, and the AMHRII-Fc fusion protein was serially diluted down the plate. All assays were performed in duplicate.

AMH ELISA

The ELISAs to detect AMH were performed similar to the AMHRII binding ELISA, except for the following modifications (41). After capturing AMH and washing as above, 50 μl/well of a 1:1000 dilution of rabbit antihuman AMH polyclonal sera was added. One hour later, the plate was washed five times with PBS/0.05% Tween 20 and 50 μl/well of a 1:5000 dilution of goat antirabbit IgG antibody conjugated to HRP (Jackson ImmunoResearch Laboratories) was added and developed as described above.

Precipitation of AMH with AMHRII-Fc-coupled Sepharose

AMHRII-Fc fusion protein (4 mg) was incubated with continuous mixing for 2.5 h at room temperature with 1 g of cyanogen bromide-activated Sepharose 4B in 2 ml of 200 mm NaCl, 100 mm sodium borate (pH. 8.4). Ethanolamine was added to 100 mm and rocked overnight at 4 C to quench the reaction. The conjugated resin was washed and stored at 4 C in PBS containing 0.02% sodium azide. Before precipitation experiments, the resin was washed three times with PBS. The AMHRII-Fc-Sepharose (20 μl) was incubated with 8 μg of AMH in 1 ml of PBS and 100 μg/ml ovalbumin for 60 min at room temperature with agitation. Tween 20 was added to 0.05%, the tube was rocked for 1 min, and the resin was recovered by centrifugation for 1 min. The resin was washed either one or more times with 1 ml of PBS, 100 μg/ml ovalbumin, and 0.05% Tween 20 for 1 min and recovered by centrifugation for 1 min. Nonreducing sample buffer (2×, 20 μl) was added, the tube was heated at 65 C for 10 min, and aliquots of the recovered supernatant were subjected to SDS-PAGE (4–20% gradient gel). C-terminal AMH was detected using an imidazole-SDS-Zn reverse staining method (51). Full-length and noncleavable AMH and pro-region were detected by Western blotting using rabbit antihuman AMH polyclonal sera (1:1000 dilution) and goat antirabbit IgG antibody conjugated to HRP (Jackson ImmunoResearch Laboratories; 1:5000 dilution). Proteins were visualized with the ECL Plus Kit Detection System (Amersham Pharmacia Biotech, Piscataway, NJ).

For experiments with biotinylated AMH, 8 μg of biotinylated-cleaved AMH were incubated with either 25 μl of Sepharose (conjugated to AMHRII-Fc- or mAb 10.6) for 4 h or 5 μg of mAb 10.6 or 22A2 for 3 h at room temperature in 1 ml of PBS containing 1% BSA, 1% goat serum. In the case of the unconjugated mAbs, antimouse Ig coupled to magnetic beads (5 μl Dynabeads Pan mouse Ig; Invitrogen) were added for the last 30 min of the incubation. The resin was recovered either by centrifugation in the case of Sepharose-conjugated reagents or with a magnet in the case of the Dynabeads and washed two times with PBS/0.05% Tween-20. Precipitated proteins were eluted from the resins as described above, subjected to SDS-PAGE (4–12% gradient gel) under reducing conditions, and detected by Western blotting using a streptavidin-HRP conjugate (Jackson ImmunoResearch Laboratories; 1:5000 dilution) and the ECL system. Images in Figure 8, A and C, were captured using a Kodak Image Station (4000MM Pro); the image in Figure 8B was captured on x-ray film. Band intensities in Figure 8C were measured using Kodak Molecular Imaging Software (version 4.5.1, Eastman-Kodak, Rochester, NY).

SMAT1 cell experiments

Smad phosphorylation was determined as previously described (38). Briefly, SMAT1 cells were seeded into six-well tissue culture plates at approximately 50% density in DMEM containing 10% fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin (Life Technologies, Gaithersburg, MD). The next day the cells were washed, medium without serum was added for 1 h, after which AMH was added in culture medium without serum for 30–90 min. The cells were washed and solubilized in 200 μl of lysis buffer [20 mm Tris-HCl, pH 7.5, 150 mm NaCl, 1% (vol/vol) Triton X-100] containing 1 mm phenylmethylsulfonyl fluoride, a proteinase inhibitor mixture (Sigma-Aldrich, St. Louis, MO), and a phosphatase inhibitor cocktail (Calbiochem Merck Biosciences, Darmstadt, Germany). The lysates were cleared by centrifugation, and supernatants were analyzed by SDS-PAGE followed by Western blotting with a rabbit anti-phospho-Smad1, 5 mAb (Cell Signaling Technology, Danvers, MA; 1:1000 dilution) and a goat antirabbit IgG antibody conjugated to HRP (Jackson ImmunoResearch Laboratories; 1:5000 dilution). The membranes were stripped and reprobed with a mouse antitubulin mAb (clone B-5-1-2, Sigma-Aldrich, St. Louis, MO).

Binding of AMH to AMHRII on COS cells

COS cells were seeded at 5 × 10⁴ cells/ml in four-well permanox Lab-Tek chambers in DMEM containing 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin (Life Technologies). Cells were transfected 24 h later with the human AMHRII cDNA using LipofectAMINE Plus reagent (Invitrogen) and cultured for 3 d. After washing with DMEM, cells were incubated for 4 h at 37 C with 3.5 nm AMH (full length, cleaved, or noncleavable) in DMEM/1% FBS, washed with DMEM, and incubated for 2 h at 37 C with 3 μg/ml of the anti-AMH mAbs in DMEM/1% FBS. After rinsing with PBS, cells were incubated 1 h with a FITC-conjugated goat antibody raised against mouse IgG in DMEM, washed with PBS, and fixed for 5 min in methanol/acetone (1/1; vol/vol). After hydration, slides were mounted in Vectashield containing propidium iodide (Vector Laboratories, Burlingame, CA) and cells were examined with a Zeiss microscope (Carl Zeiss, Jena, Germany).

Detection of AMH in COS cells

COS cells were transfected with the human AMH cDNA as described above. Cells were washed 3 d later with PBS and fixed and permeabilized by incubation in methanol/acetone (1/1; vol/vol) for 5 min. After blocking with DMEM/1% FBS for 1 h, cells were incubated for 2 h at 37 C with 3 μg/ml anti-AMH mAbs in DMEM/1% FBS medium, washed with DMEM, and incubated 1 h with a FITC-conjugated goat antibody raised against mouse IgG in DMEM. After a wash with PBS, slides were mounted and examined as described above.