Inhibin is a heterodimeric TGFβ family ligand that is expressed in many cancers and is a selective biomarker for ovarian cancers; however, its tumor-specific functions remain unknown. Here, we demonstrate that the α subunit of inhibin (INHA), which is critical for the functionality of dimeric inhibin A/B, correlates with microvessel density in human ovarian tissues and is predictive of poor clinical outcomes in multiple cancers. We demonstrate that inhibin-regulated angiogenesis is necessary for metastasis. Although inhibin had no direct impact on tumor cell signaling, both tumor cell-derived and recombinant inhibin elicit a strong paracrine response from endothelial cells by triggering SMAD1/5 activation and angiogenesis in vitro and in vivo. Inhibin-induced angiogenesis was abrogated via anti-inhibin α antibodies. The endothelial-specific TGFβ receptor complex comprising ALK1 and endoglin was a crucial mediator of inhibin signaling, offering a molecular mechanism for inhibin-mediated angiogenesis. These results are the first to define a role for inhibin in tumor metastasis and vascularization and offer an antibody-based approach for targeting inhibin therapeutically.

Significance: Inhibin is a predictor of poor patient survival in multiple cancers and is a potential target for antiangiogenic therapies. Cancer Res; 78(11); 2978–89. ©2018 AACR.

Inhibition of angiogenesis, the growth of new blood vessels from preexisting vasculature, is a clinically validated anticancer strategy for numerous tumor types. However, although the VEGF/VEGF receptor (VEGFR) signaling axis is widely recognized as the principal target of this therapeutic approach, current FDA-approved anti-VEGF drugs have demonstrated sub-optimal responses in the clinic. In many cases, including in ovarian cancers (OVCA), patient relapse, acquired resistance and cytotoxicity is commonly observed following anti-VEGF therapy. The identification of new vascular targets with potentially fewer adverse effects is therefore critical for improving therapeutic outcomes.

TGFβ family members, particularly TGFβ1 and BMP9, are essential regulators of angiogenesis (1). Targeting these for antiangiogenic therapy remains a formidable challenge due to their nonendothelial pleiotropic functions. Here, we focus on the unique TGFβ family member inhibin, an endocrine hormone that sharply declines at the onset of menopause in healthy normal women and remains low (2) unlike other TGFβ family members and prototypical angiogenic factors like VEGF. Importantly, when inhibin becomes elevated in postmenopausal women, this elevation becomes a diagnostic and prognostic marker for OVCA where along with CA125 detects 95% of ovarian tumors with 95% specificity (3, 4).

Inhibin is a heterodimeric member of the TGFβ family composed of an alpha (α; coded by INHA) and β subunit (INHBA or INHBB). Combinations of these subunits give rise to either inhibin A (αβA) or inhibin B (αβB; 5). Although inhibin null mice (INHA−/−) are viable, they present with alterations to the ovarian vasculature and result in spontaneous gonadal tumors (6). In humans, however, inhibin levels are elevated in multiple cancer types, including ovarian, prostate, adrenal, stomach, and pancreatic cancers with indications for a role for inhibin in prostate cancer metastasis (7–11). Despite these findings, the functional consequences of elevated tumor-derived inhibin have yet to be determined.

Several inhibin-binding proteins/receptors were previously reported (12). However, unlike other TGFβ members, whose signal transduction mechanisms have been well studied, the mechanisms of inhibin signaling remain largely unclear. The best-characterized inhibin-binding protein is the epithelial cell surface TGFβ coreceptor TβRIII/betaglycan (13). Inhibin binding to betaglycan fails to activate any discernable downstream pathways in epithelial cells. Others and we previously demonstrated, several tumor suppressor functions for betaglycan, which is lost in the majority of human cancers (14), but little is known about the impact of elevated inhibin on nonepithelial cells that do not express significant betaglycan.

Given the urgent need to identify new antiangiogenic pathways and targets to complement and improve existing therapies, we examined the potential role of inhibin as a novel regulator of angiogenesis and metastasis. Importantly, we demonstrate inhibin as an unexpected, clinically relevant, paracrine factor of tumor-induced angiogenesis and define the underlying mechanism of inhibin action and therapeutic potential.

Cell lines and reagents

Ovarian epithelial carcinoma cell lines were obtained either from Duke Gynecology/Oncology Bank (Durham, NC) and ATCC. Authentication was carried out at the University of Colorado (Denver, CO) sequencing facility. HMEC-1 (human dermal microvascular endothelial cells) from ATCC CRL-3243 and murine embryonic endothelial cells (MEEC) ENG+/+ and ENG−/− were as described previously (15). HUVEC (human umbilical vein endothelial cells) was purchased from Lonza. HMEC-1s were grown as per the ATCC instructions. Epithelial carcinoma cell lines A2780, HEY, IGROV, OVCA247, M41, OVCA3, OVCA4, OVCA420, OVCA429, OVCA448, SKOV3 and PA1 were cultured in RPMI-1640 (ATCC 30–2001) containing l-glutamine, 10% FBS and 100 U of penicillin–streptomycin. All cells lines were maintained at 37°C in a humidified incubator at 5% CO2, routinely checked for Mycoplasma 3 times a year and experiments conducted within 3–6 passages depending on the cell line. Antibodies phospho-SMAD1/5 (#9516), phospho-SMAD2/3 (#8828S) and SMAD2/3 (#5678S) were from Cell Signaling Technology, SMAD1/5 (#ab75273) from Abcam. Mouse anti-HA antibody, rabbit anti-HA antibody and mouse anti-Myc antibody were from invitrogen. Monoclonal antibodies to CD31 (#F8402) and inhibin α (#ab47720) for IHC were purchased from Sigma-Aldrich and Abcam, respectively. Anti-INHA antibody (polyclonal #sc22048, Santa Cruz Biotechnology) and (monoclonal #sc365439, Santa Cruz Biotechnology) were used as indicated. ML347, Dorsomorphin and SB351432 were from Sigma-Aldrich, TRC105 was a gift from TRACON pharmaceuticals (http://www.traconpharma.com/trc105.php). Inhibin A was from Sigma-Aldrich (# I9149) and R&D Systems (# 8506-AB). Lentiviral particles were generated at the COBRE Center for Targeted Therapeutics Core Facility at South Carolina. For INHA knockdown, SKOV3 cells were infected with shRNA lentivirus, selected in 2 μg/mL puromycin and stable cell lines maintained in 1 μg/mL puromycin. Transient DNA transfections of HMEC-1 and COS7 were performed using either Targetfect (#HUVEC-01) from Targeting Systems or Lipofectamine 2000 (#11668019) from Life Technologies.

RNA isolation and qRT-PCR analysis

Total RNAs was extracted using TRIzol and chloroform. RNA was retro-transcribed using iScript Reverse Transcription Supermix (#1708841) and Advanced Universal SYBR Green Supermix (#1725271) from Bio-Rad. All expression data were normalized to those for RPL13A. qRT-PCR primer sequences are listed in Table 1.

Table 1.

qRT-PCR primer sequences

PrimersForwardReverse
Human   
RPL13A AGATGGCGGAGGTGCAG GGCCCAGCAGTACCTGTTTA 
ENG CGCCAACCACAACATGCAG GCTCCACGAAGGATGCCAC 
INHA TTCCACTACTGTCATGGTGGT AGTGCTGCGTGAGAAGGTTG 
ALK1 CATCGCCTCAGACATGACCTC GTTTGCCCTGTGTACCGAAGA 
ALK2 GTGAAGGTCTCTCCTGCGGTA GCCATCGTTGATGCTCAGTGA 
ALK3 TGAAATCAGACTCCGACCAGA TGGCAAAGCAATGTCCATTAGTT 
ALK4 CTTCCCCCTTGTTGTCCTCC TCTCACACGTGTAGTTGGCC 
ALK5 TCCCAAACAGATGGCAGAGC ATCCGCAATGCTGTAAGCCT 
ALK6 ATGGAACTTGCTGTATTGCT CAACTCGAGTGTTAGGTGGT 
ALK7 GATGTGACCGCCTCTGGATC CCATCTTCCATGCCACACCT 
Mouse   
RPL13A CAAGGTTGTTCGGCTGAAGC GCTGTCACTGCCTGGTACTT 
ENG GCTGAAGACACTGACGACCA AGCCTGACGGGAAACTGATG 
PrimersForwardReverse
Human   
RPL13A AGATGGCGGAGGTGCAG GGCCCAGCAGTACCTGTTTA 
ENG CGCCAACCACAACATGCAG GCTCCACGAAGGATGCCAC 
INHA TTCCACTACTGTCATGGTGGT AGTGCTGCGTGAGAAGGTTG 
ALK1 CATCGCCTCAGACATGACCTC GTTTGCCCTGTGTACCGAAGA 
ALK2 GTGAAGGTCTCTCCTGCGGTA GCCATCGTTGATGCTCAGTGA 
ALK3 TGAAATCAGACTCCGACCAGA TGGCAAAGCAATGTCCATTAGTT 
ALK4 CTTCCCCCTTGTTGTCCTCC TCTCACACGTGTAGTTGGCC 
ALK5 TCCCAAACAGATGGCAGAGC ATCCGCAATGCTGTAAGCCT 
ALK6 ATGGAACTTGCTGTATTGCT CAACTCGAGTGTTAGGTGGT 
ALK7 GATGTGACCGCCTCTGGATC CCATCTTCCATGCCACACCT 
Mouse   
RPL13A CAAGGTTGTTCGGCTGAAGC GCTGTCACTGCCTGGTACTT 
ENG GCTGAAGACACTGACGACCA AGCCTGACGGGAAACTGATG 

Plasmid constructs and stable cell lines

The shRNA sequences for INHA were obtained either from Sigma-Aldrich or Dharmacon. ShINHA1 used in all the experiments was generated using TRCN0000063904: CCGGCCTCGGATGGAGGTTACTCTTCTCGAGAAGAGTAACCTCCATCCGAGGTTTTG in a pLKO1-puromycin backbone. Scramble vector was used as control (nontargeted control). The second shRNA sequences from Dharmacon: shRNA V3SH11240-226731666: TTCAGTGCTACTCTGTGGC. shRNA for ALK1 was obtained from Sigma-Aldrich TRCN0000000356: CCGGCAGTCCAGAGAAGCCTAAAGTCTCGAGACTTTAGGCTTCTCTGGACTGTTTTT. Human endoglin shRNA was described previously (16).

ELISA

The enzyme linked immunosorbent assay (ELISA) was from AnshLabs (#AL-134) and was performed according to the manufacturer's instructions to quantitatively measure total inhibin (inhibin A, inhibin B and inhibin α; alpha). Cells were serum starved for 24 hours before conditioned media (CM) was collected.

Co-patch immunofluorescence

Antibody-mediated immunofluorescence co-patching was performed as described previously (17, 18). HA and Myc tags at the cell surface were immunolabeled at 4°C for 45 minutes using anti-HA and anti-Myc antibodies, to allow only surface labeling and to reduce internalization. Alexa 488 and Alexa 594 conjugated to anti-HA and anti-Myc were used as secondary antibodies, respectively. Imaging was performed using an Olympus IX81 inverted microscope (Shinjuku). Images were exported to ImageJ. The number of superimposed (red and green) yellow patches were counted manually on flat cell regions as described previously (17, 18). >50 patches/cell were counted.

Endothelial capillary sprouting and tube formation assay

Tube formation in HMEC-1, HUVEC, and MEECs was performed as described previously (16, 19) using endothelial cells plated between two layers of Matrigel except in the case of MEEC's, which were plated on a single layer of Matrigel. Briefly, 24-well plates were coated with 200 μL of Matrigel matrix (#354230, BD Biosciences). Thirty minutes following the addition of Matrigel, HMEC-1 cells (70 × 103), HUVEC (50 × 103) or MEEC (50 × 103) were plated onto the matrix. For HMEC-1′s, a second layer of 200 μL Matrigel was added after 1 hour of plating cells. After a 30 minutes incubation at 37°C, 300 μL growth medium with growth factors (300 pmol/L inhibin A, TGFβ, BMP9, or activin as indicated) or CM collected from shControl or shINHA stable cell lines as indicated was added. Sixteen hours later images were taken using an Olympus IX81 inverted microscope (×4 magnification).

For all tube formation experiments, quantification presented represents the two commonly used angiogenesis parameters in vitro (20): number of meshes that are defined as regions enclosed by segments (tubes delimited by two junctions) and/or closed polygons; nodes/branches, which are defined as individual junctions/branching points. The Angiogenesis Analyzer plugin in ImageJ was used for the analysis. For spheroid-based sprout assays, endothelial spheroids were prepared as previously reported (21). Briefly, 1 × 103 HMEC-1 cells were cultured in hanging drops of 25 μL medium containing 20% methocel and 80% culture medium, and allowed to aggregate as spheroids. After 24 hours, the spheroids were collected and plated on 24-well plates coated with growth factor–reduced Matrigel and treated as indicated. Sprouts were digitally imaged after the indicated times and the number and length of sprouts per spheroid quantitated. For all experiments, a minimum of two biological trials, with each trial containing three technical replicates were analyzed by counting a minimum of 3 fields/technical replicate.

Study approval

All animal experimental protocols were performed in accordance with the Institutional Animal Care and Use Committee at the University of South Carolina under an approved protocol (AUP 2329-101161-121916).

In vivo Matrigel plug assay

Matrigel plug assays were carried out by using Matrigel (BD Biosciences, #354230) mixed with 100 ng/mL inhibin A or mixed with CM from shControl/shINHA SKOV3 in a ratio of 2:3 (total volume of 0.2 mL) and injected subcutaneously into the right flank of BALB/c female mice ages 5 to 6 weeks. n = 3 mice per group were used (22, 23) with each experiment conducted two independent times. Plugs were harvested 12 days after injection and hemoglobin content was determined according to the Drabkin's method.

Orthotopic xenografts

Stable 5 × 106 of shControl or shINHA SKOV3 cells were intraperitoneally injected into 7-week-old female NCr nude mice (Taconic Biosciences) athymic nude mice that were housed under pathogen-free conditions on a 12-hour-light/dark cycle. Animals were monitored closely for tumor growth and signs of illness and sacrificed at humane end points. Cells were pathogen free and suspended in 200 μL HBSS. Two independent biological trials were conducted (first with 5 mice per group and the second with 8 mice per group) and monitored weekly with girth measurements in triplicates. At 7 weeks (or if mice lost 20% of body weight), incisions were made in the mice abdomen and ascites collected, followed by necropsy for metastatic burden quantitation.

DAB-based IHC

The tissue microarray (TMA) of OVCA tumor cores was purchased from Protein Biotechnologies Inc. (#TMA-2213). The formalin-fixed, paraffin-embedded tissue arrays were deparaffinized by sequential washing with xylene, 100% ethanol, 90% ethanol, 80% ethanol, 70% ethanol and distilled water for 10 minutes each. For antigen retrieval the tissue sections were heated for 10 minutes in Sodium citrate buffer (pH 6.0) or Tris EDTA buffer [pH 9.0; anti-INHA (inhibin α) or anti-CD31, respectively]. Endogenous peroxidases were blocked with 3% H2O2 for 10 minutes, followed by block with 10% normal goat serum in TBST for 1 hour room temperature, followed by incubation with primary antibody overnight at 4°C in a humidified chamber. Appropriate secondary antibody conjugated to horseradish peroxidase in blocking solution was added for 30 minutes at room temperature. Horseradish peroxidase was detected with 3,3′-diaminobenzidine (DAB; ThermoFisher Scientific) substrate for 5 minutes, washed and counterstained with hematoxylin. Monoclonal antibodies to CD31 (#F8402, Sigma) and inhibin α (#ab47720) from Abcam were used. Stained specimens were examined using a phase-contrast microscope (model IX70; Olympus). For inhibin α levels, semiquantitative analysis was performed independently by 2 blind investigators (one pathologist) using a 3-tiered scoring system (none/trace/low, medium, or high being dark to high staining). Discrepancies between the 2 investigators were discussed and reconciled (<10 samples). For CD31 analysis, images were processed using ImageJ (NIH) to quantify CD31-positive areas. To quantify microvessel density (MVD), microvessel-like structures consisting of endothelial cells that were stained with anti-CD31 antibody were counted in similar fields in the entire core by setting a constant threshold and presented as number of vessels/mm2 of tissue section. Five random fields of each section were analyzed.

Statistical analysis

All clinical and xenograft data were analyzed using nonparametric statistics. Survival curves were analyzed with log-rank statistics. In vitro experiments were analyzed using parametric statistics (ANOVA global test with Bonferroni-corrected two-tailed Student t tests as post-hoc tests) and presented as mean ± SEM. In cases where data were normalized to control, 1-sample Student t test was used with an expected value of 1% or 100% to decrease the likelihood of a type I error. All statistical analyses were conducted with GraphPad Prism Software.

Inhibin α analysis in ovarian cancer and impact on patient survival

Prior reports on analysis of inhibin α (encoded by INHA) in tumors, have indicated INHA and inhibin α in circulation, including its utility as a biomarker for ovarian cancers (OVCA). To assess parameters in tumor tissues that could reveal inhibin function in tumors, we first evaluated inhibin α expression in human normal and primary OVCA tissues using immunohistochemistry on an ovarian TMA with 75 distinct tissue cores spanning different OVCA subtypes. We find that, whereas inhibin α was detectable at low levels in the tissue cores from normal patients, 46 % of endometrial, 51 % of serous and 44 % of mucinous tumors expressed high inhibin α (Fig. 1A and B; Supplementary Fig. S1, i and S1, ii), suggesting a broad increase in inhibin α across all OVCA subtypes. The levels of inhibin α were further elevated in higher tumor grades in all subtypes (Fig. 1B, ii). These results are in agreement with prior reports. Using this TMA, we noted inhibin α localization exclusively in epithelial cells in 20% of tumors and in both epithelial and stromal cells in the remaining 80% of tumors (Supplementary Table S1), suggesting potential stromal inhibin functions. We next examined changes in MVD using CD31 staining in the same tissues. A total of 52 cores were scorable for both inhibin α and CD31 based on quality of cores. We find that tumor cores that expressed either medium and/or high levels of inhibin α (inhibin αmed/high, N = 29) exhibited 2.5 times more microvessels per mm2 as compared with inhibin αlow tumor cores (N = 23; Fig. 1C; Supplementary Fig. S1, iii). Regression analysis revealed that higher MVD correlated with higher inhibin α levels, with a Spearman correlation of 0.337(P < 0.05).

Figure 1.

Inhibin α expression in ovarian carcinoma and RCC and association with angiogenesis and overall survival. A, Representative images from IHC from a human ovarian cancer TMA of either normal tissue or different ovarian cancer subtypes with high inhibin α staining as determined by immunolabeling with anti-inhibin α antibody and IgG control (Supplementary Fig. S1, i). Top, representative high staining cores across subtypes as shown. Zoomed in examples are in Supplementary Fig. S1, ii. Bottom, representative cores scored using a three-tier system (Materials and Methods) as none/trace/low, medium, or high staining as shown across subtypes. B, Quantitation represented as the percentage of cores with inhibin α immunoreactivity scored as none/trace/low, medium, or high staining for (i) each subtype and (ii) relative to tumor grade (I–III) as indicated. C, IHC of tissue arrays immunolabeled with anti-inhibin α or anti-CD31 antibody. Representative images of CD31 staining with the corresponding inhibin α levels from the same cores. Additional examples are in Supplementary Fig. S1, iii. Chart represents MVD quantitation in tumor cores with respect to inhibin α levels quantified as in A and B and described in Materials and Methods. **, P < 0.01 (Mann–Whitney test). Medium and high cores were combined as indicated. D, Overall patient survival with low or high inhibin α (INHA) mRNA in either ovarian cancer or renal clear cell carcinomas. E, INHA mRNA expression by qRT-PCR to detect endogenous INHA mRNA in a panel of ovarian cancer cell lines and endothelial cells as indicated.

Figure 1.

Inhibin α expression in ovarian carcinoma and RCC and association with angiogenesis and overall survival. A, Representative images from IHC from a human ovarian cancer TMA of either normal tissue or different ovarian cancer subtypes with high inhibin α staining as determined by immunolabeling with anti-inhibin α antibody and IgG control (Supplementary Fig. S1, i). Top, representative high staining cores across subtypes as shown. Zoomed in examples are in Supplementary Fig. S1, ii. Bottom, representative cores scored using a three-tier system (Materials and Methods) as none/trace/low, medium, or high staining as shown across subtypes. B, Quantitation represented as the percentage of cores with inhibin α immunoreactivity scored as none/trace/low, medium, or high staining for (i) each subtype and (ii) relative to tumor grade (I–III) as indicated. C, IHC of tissue arrays immunolabeled with anti-inhibin α or anti-CD31 antibody. Representative images of CD31 staining with the corresponding inhibin α levels from the same cores. Additional examples are in Supplementary Fig. S1, iii. Chart represents MVD quantitation in tumor cores with respect to inhibin α levels quantified as in A and B and described in Materials and Methods. **, P < 0.01 (Mann–Whitney test). Medium and high cores were combined as indicated. D, Overall patient survival with low or high inhibin α (INHA) mRNA in either ovarian cancer or renal clear cell carcinomas. E, INHA mRNA expression by qRT-PCR to detect endogenous INHA mRNA in a panel of ovarian cancer cell lines and endothelial cells as indicated.

Close modal

Using Cox proportional hazards regression to analyze the impact of INHA expression on patient survival, we find that higher inhibin α mRNA (INHA) was associated with shorter overall survival (OS) in TP53 mutated OVCA's (P < 0.05; HR, 1.42, Fig. 1D). Because INHA levels correlate with CD31 in ovarian tumors, we examined whether INHA level in tumors would be a stronger predictor of survival in renal clear cell carcinomas that are highly angiogenic and for which antiangiogenic therapies are the mainstay of treatment. We find that renal clear cell carcinoma patients with high INHA had significantly worse OS, with a median OS of 91.7 versus 54.2 months for tumors with low and high INHA, respectively (P < 0.005; HR, 2.02, Fig. 1D). On the basis of these observations, to examine whether INHA expression and inhibin α secretion by tumors was causal to the increased vascular density and to identify model systems for examining the role of inhibin in tumors, we first examined INHA mRNA levels across a panel of OVCA and endothelial cells. We find a spectrum of secreted inhibin α and INHA mRNA expression in epithelial cells (Supplementary Fig. S1, iv; Fig. 1E) in contrast with low to no INHA expression in all four commonly used-endothelial cell lines tested (Fig. 1E). Although these data do not rule out endothelial cells as potential sources of INHA in the tumor, they suggest epithelial cells as the likely source of inhibin in OVCA tissues, and point to a potential correlative impact of inhibin α expression on tumor vasculature.

Tumor cell–produced inhibin α and recombinant inhibin A induce specifically endothelial cell responses

Given the correlation between inhibin α expression and MVD in tumors (Fig. 1C), we examined the impact of epithelial-produced inhibin on endothelial cell biology by, using shRNA-mediated suppression of INHA mRNA in epithelial cancer cells. We used SKOV3 cancer cells (high INHA mRNA), and two independent shRNAs to reduce INHA mRNA (Fig. 2A; Supplementary Fig. S2: 2nd shRNA). Reduction in INHA mRNA correlated with reduced secretion of inhibin α and total dimeric inhibin (A/B; from here on referred to as inhibin), as confirmed using an ELISA to detect total inhibin (Fig. 2A, ii). We find that endothelial cells incubated with CM from scramble PLKO1 shRNA (shControl) cells robustly induced tube formation of endothelial cells 4.5 times more than CM from shINHA cells (Fig. 2B), which secrete less inhibin (Fig. 2A, ii). This finding was confirmed using CM from a second independent shRNA to INHA (Supplementary Fig. S2, i). Because TGFβ members besides inhibin could be altered in shINHA SKOV3 cells, we aimed to determine whether the CM effects were specific to inhibin by examining endothelial cell tubulogenesis in the presence of an anti-inhibin α antibody. We find that incubating CM with two independent anti-inhibin α antibodies (polyclonal and monoclonal) suppresses CM-induced tube formation 2.5 times more than controls (Fig. 2B; Supplementary Fig. S2, ii). The effect of anti-inhibin α was not tumor cell line specific as anti-inhibin α also suppressed CM induced tube formation from a second OVCA cell line M41 (Supplementary Fig. S2, iii). These findings are the first to demonstrate the use of an anti-inhibin α antibody in the suppression of any tumor promoting phenotype. We next used recombinant inhibin (inhibin A) to examine inhibin potency in tube formation alongside equimolar amounts of other TGFβ members that are known angiogenesis regulators: BMP9, activin, and TGFβ, which either suppress or promote angiogenesis in different contexts (24, 25). We find that inhibin induced tube formation 9 and 4.5 times more effectively than TGFβ or activin, respectively, and to a similar extent as BMP9 (Fig. 2C). Combining inhibin with either activin or TGFβ resulted in a significant increase in tube formation with no further increase observed upon combining inhibin with BMP9 (Fig. 2C). A second commercial source of inhibin (Supplementary Fig. S3, i) and an antibody to block recombinant inhibin (Supplementary Fig. S3, ii) confirmed the specificity of inhibin's impact on tube formation in multiple endothelial cell types (Fig. 2D). To examine whether the response of endothelilal cells to inhibin was restricted to tube formation on Matrigel, we tested chemotactic migration toward inhibin, endothelial cell spheroid sprouting (Materials and Methods), and cell proliferation. No effect of inhibin on endothelial cell proliferation was observed (Supplementary Fig. S4i). However, inhibin increased endothelial cell migration toward recombinant inhibin and increased endothelial spheroid sprouting (Supplementary Fig. S4, ii and S4, iii). Notably, no effect of shRNA to inhibin was observed on tumor cell growth (Supplementary Fig. S4, iv) or Matrigel invasion in vitro (Supplementary Fig. S4, v). These data suggest a likely endothelial-specific response to inhibin.

Figure 2.

Inhibin increases endothelial cell tubulogenesis. A, (i) INHA mRNA levels in stable SKOV3 shControl and shINHA cancer cells (Materials and Methods). (ii) ELISA for total inhibin (inhibin α, inhibin A/B) from shControl and shINHA SKOV3 CM. B, Tube formation of HMEC-1 cells in the presence of CM from shControl SKOV3 cancer cells or shINHA SKOV3 cells (i) and 10 μg/mL anti-inhibin α or IgG control (ii) or treated with 300 pmol/L recombinant inhibin A alone or in the presence of anti-inhibin α antibody [10 μg/mL; (iii)]. Bar graphs represent the average number of meshes quantified and represents an average of independent trials (Materials and Methods). Mean ± SEM; ***, P < 0.00 and **, P < 0.01, Student t test; scale bar, 500 μm. C, HMEC-1 treated with equimolar amounts (300 pmol/L) of indicated TGFβ members either alone or in combination with inhibin A. Bar graph represents quantitation as in B. One-way ANOVA, Tukey multiple comparison test, ***, P < 0.001 and **, P < 0.01. D, Tube formation of HMEC-1, HUVEC, and MEEC cells upon treatment with 300 pmol/L inhibin A for 16 hours. Graph represents number of meshes, branches, and nodes (Materials and Methods). Mean ± SEM; ***, P < 0.001 and **, P < 0.01, Student t test; scale bar, 500 μm.

Figure 2.

Inhibin increases endothelial cell tubulogenesis. A, (i) INHA mRNA levels in stable SKOV3 shControl and shINHA cancer cells (Materials and Methods). (ii) ELISA for total inhibin (inhibin α, inhibin A/B) from shControl and shINHA SKOV3 CM. B, Tube formation of HMEC-1 cells in the presence of CM from shControl SKOV3 cancer cells or shINHA SKOV3 cells (i) and 10 μg/mL anti-inhibin α or IgG control (ii) or treated with 300 pmol/L recombinant inhibin A alone or in the presence of anti-inhibin α antibody [10 μg/mL; (iii)]. Bar graphs represent the average number of meshes quantified and represents an average of independent trials (Materials and Methods). Mean ± SEM; ***, P < 0.00 and **, P < 0.01, Student t test; scale bar, 500 μm. C, HMEC-1 treated with equimolar amounts (300 pmol/L) of indicated TGFβ members either alone or in combination with inhibin A. Bar graph represents quantitation as in B. One-way ANOVA, Tukey multiple comparison test, ***, P < 0.001 and **, P < 0.01. D, Tube formation of HMEC-1, HUVEC, and MEEC cells upon treatment with 300 pmol/L inhibin A for 16 hours. Graph represents number of meshes, branches, and nodes (Materials and Methods). Mean ± SEM; ***, P < 0.001 and **, P < 0.01, Student t test; scale bar, 500 μm.

Close modal

Effect of inhibin on tumor angiogenesis and metastasis

Given the SKOV3-derived and recombinant inhibin-dependent effects on tube formation in vitro (Fig. 2), we examined the impact of paracrine produced and recombinant inhibin in an in vivo Matrigel plug assay. Compared with endothelial cell growth supplement (ECGS, positive control) and CM from shControl SKOV3 cells (Fig. 3A), CM from shINHA SKOV3 cells that secrete less inhibin (Fig. 2A) had substantially reduced functional blood flow into the Matrigel plug, as measured using hemoglobin content as a surrogate marker (Fig. 3A, 40% decrease, **, P < 0.01; refs. 23, 26). Because components in CM besides inhibin could impact angiogenesis into the Matrigel plug in vivo, we next examined whether recombinant inhibin alone was sufficient to increase blood flow in the Matrigel plug assay using 100 ng/mL ECGS as a positive control. We find that compared with PBS, 100 ng/mL inhibin led to twice as much hemoglobin content with visibly distinct vasculature in the Matrigel plug (Fig. 3B, 50% increase relative to PBS, *, P < 0.05). These data together indicate that inhibin induces blood vessel formation in vivo.

Figure 3.

Inhibin increases blood vessel formation, tumor angiogenesis, and metastatic burden in vivo. Representative images (top) and hemoglobin quantification using Drabkins reagent (bottom) of Matrigel plugs 12 days post-subcutaneous injection of growth factor–reduced Matrigel mixed with either 100 ng/mL ECGS (positive control), shControlCM, or shINHA CM from SKOV3-stable cells (A) or recombinant inhibin A (100 ng/mL) or PBS (negative control; B). Mean ± SEM; ***, P < 0.001; **, P < 0.01; and *, P < 0.05, ANOVA with Holm–Sidak post hoc test; n = 3 mice per condition; scale bar, 20 μm. C, Nude mice were injected intraperitoneally with 5 × 106 shControl or shINHA SKOV3 cells, animals were sacrificed at 7 weeks and/or if body weight loss reached 20% as a humane endpoint. Ascites and metastatic burden were examined at necropsy: (i) number of mice per group that had ascites, (ii) abdominal girth measurements (in triplicate) at the beginning and end points, (iii) overall weight of tumors/lesions found in the peritoneal cavity. *, P < 0.05, χ2 test for (i) and Mann–Whitney U test for (ii) and (iii). n = 13 mice per group (two independent trials combined). Sample mouse images are provided in Supplementary Fig. S4, vi.

Figure 3.

Inhibin increases blood vessel formation, tumor angiogenesis, and metastatic burden in vivo. Representative images (top) and hemoglobin quantification using Drabkins reagent (bottom) of Matrigel plugs 12 days post-subcutaneous injection of growth factor–reduced Matrigel mixed with either 100 ng/mL ECGS (positive control), shControlCM, or shINHA CM from SKOV3-stable cells (A) or recombinant inhibin A (100 ng/mL) or PBS (negative control; B). Mean ± SEM; ***, P < 0.001; **, P < 0.01; and *, P < 0.05, ANOVA with Holm–Sidak post hoc test; n = 3 mice per condition; scale bar, 20 μm. C, Nude mice were injected intraperitoneally with 5 × 106 shControl or shINHA SKOV3 cells, animals were sacrificed at 7 weeks and/or if body weight loss reached 20% as a humane endpoint. Ascites and metastatic burden were examined at necropsy: (i) number of mice per group that had ascites, (ii) abdominal girth measurements (in triplicate) at the beginning and end points, (iii) overall weight of tumors/lesions found in the peritoneal cavity. *, P < 0.05, χ2 test for (i) and Mann–Whitney U test for (ii) and (iii). n = 13 mice per group (two independent trials combined). Sample mouse images are provided in Supplementary Fig. S4, vi.

Close modal

Angiogenesis is critical for peritoneal metastasis manifested by accumulation of ascites in >70% of advanced human OVCA (27, 28). To mimic the widespread intra-abdominal metastasis observed in the peritoneal cavity during OVCA progression (29, 30), we injected nude mice intraperitoneally (intrapertoneal injection) and used this in vivo model to determine whether the reduced angiogenesis observed in shINHA SKOV3 cells in Matrigel plugs and in in vitro analysis (Fig. 2) translated to alterations in peritoneal burden and ascites accumulation. We intrapertoneally injected either shControl or shINHA SKOV3 cells into 7-week-old athymic female nude mice. Seven weeks after tumor cell injections, mice were sacrificed and ascites and tumor burden quantified (Materials and Methods). We find that all mice contained ovarian tumors (Supplementary Fig. S4, vi, dotted yellow circles). However, while 10/13 mice injected with shControl SKOV3 cells accumulated ascites, only 5/13 mice injected with shINHA cells had ascites (Fig. 3C, i). Consistently, only mice receiving shControl SKOV3 cells showed an increase in abdominal girth (initial girth vs. final girth; Fig. 3C, ii). Necropsy confirmed that tumor burden in the abdominal cavity was significantly higher in mice receiving shControl cells as compared with mice receiving shINHA SKOV3 cells (Fig. 3C, iii). Together, these findings are the first to link a reduction in inhibin with suppression of angiogenesis and metastatic burden in vivo.

Inhibin specifically promotes SMAD1/5 activation in endothelial cells using the TGFβ coreceptor endoglin

Because TGFβ family members transduce signals via SMAD2/3 or SMAD1/5 pathways, we examined the effects of paracrine and recombinant inhibin on SMAD2/3 and SMAD1/5 phosphorylation in endothelial cells. We find that CM from shControl SKOV3 cells (Fig. 2A) specifically increased phosphorylation of SMAD1/5 in endothelial cells (Fig. 4A), whereas no change in phosphorylation of SMAD2/3 was observed (Fig. 4A). Reducing inhibin in the CM of SKOV3 cells by INHA shRNA (Fig. 2A) resulted in significantly reduced SMAD1/5 activation in endothelial cell lines (Fig. 4A). We further tested the specificity of inhibin's effects within CM on SMAD1/5 phosphorylation by incubating CM with an inhibin α antibody. We find that CM-induced SMAD1/5 phosphorylation was suppressed significantly by the antibody to similar extents as shINHA SKOV3 CM (Supplementary Fig. S5). To determine the specificity and kinetics of the SMAD1/5 activation, we used recombinant inhibin and find that recombinant inhibin robustly activated only SMAD1/5, and not SMAD2/3 (Fig. 4B) in multiple endothelial cells in a time and dose dependent manner (Supplementary Fig. S6, i). In contrast with the signaling responsiveness of endothelial cells to recombinant inhibin, SKOV3 epithelial cancer cells did not stimulate phosphorylation of SMAD1/5 in response to recombinant inhibin (Supplementary Fig. S6, ii).

Figure 4.

Endoglin is essential for inhibin-induced signaling and tubulogenesis. A, Western blot analysis of lysates from HMEC-1 cells serum starved for 6 hours and then incubated with indicated CM for up to 60 minutes, followed by immunoblotting of lysates. Quantitation of pSMAD1/5 changes (right graph) from two independent biological trials B, Western blot analysis of lysates from HMEC-1 or MEECs serum starved for 6 hours and then incubated with 20 pmol/L inhibin A for up to 30 minutes, followed by immunoblotting of lysates as indicated. Quantitation of pSMAD1/5 changes (right graph) from two independent biological trials. C, Tube formation of HMEC-1 cells in the presence or absence of recombinant inhibin A (300 pmol/L) either untreated or pretreated with 100 μg/mL TRC105 for 30 minutes before inhibin A treatment. Average number of meshes/field quantified 16 hours after inhibin treatment (Materials and Methods). Scale bar, 500 μm. D, qRT-PCR analysis of endoglin mRNA after transient knockdown of endoglin using either control shRNA in HMEC-1 (shControl) or shRNA to endoglin, followed by (i) tube formation in the absence or presence of recombinant inhibin A. Graph of average number of meshes as described in Materials and Methods; scale bar, 500 μm. (ii) Western blotting for SMAD1/5 phosphorylation in response to 20 pmol/L inhibin A for up to 60 minutes. Quantitation of pSMAD1/5 changes (right graph) from two independent biological trials is presented. E, (i) Tube formation in MEEC ENG+/+ cells and endoglin null ENG−/− MEECs cells in the absence or presence of inhibin A after 16 hours. Graph of average number of meshes as described in Materials and Methods; scale bar, 500 μm. (ii) Western blotting for SMAD1/5 phosphorylation in response to 20 pmol/L inhibin A for up to 60 minutes in MEEC ENG+/+ or ENG−/− cells as indicated.

Figure 4.

Endoglin is essential for inhibin-induced signaling and tubulogenesis. A, Western blot analysis of lysates from HMEC-1 cells serum starved for 6 hours and then incubated with indicated CM for up to 60 minutes, followed by immunoblotting of lysates. Quantitation of pSMAD1/5 changes (right graph) from two independent biological trials B, Western blot analysis of lysates from HMEC-1 or MEECs serum starved for 6 hours and then incubated with 20 pmol/L inhibin A for up to 30 minutes, followed by immunoblotting of lysates as indicated. Quantitation of pSMAD1/5 changes (right graph) from two independent biological trials. C, Tube formation of HMEC-1 cells in the presence or absence of recombinant inhibin A (300 pmol/L) either untreated or pretreated with 100 μg/mL TRC105 for 30 minutes before inhibin A treatment. Average number of meshes/field quantified 16 hours after inhibin treatment (Materials and Methods). Scale bar, 500 μm. D, qRT-PCR analysis of endoglin mRNA after transient knockdown of endoglin using either control shRNA in HMEC-1 (shControl) or shRNA to endoglin, followed by (i) tube formation in the absence or presence of recombinant inhibin A. Graph of average number of meshes as described in Materials and Methods; scale bar, 500 μm. (ii) Western blotting for SMAD1/5 phosphorylation in response to 20 pmol/L inhibin A for up to 60 minutes. Quantitation of pSMAD1/5 changes (right graph) from two independent biological trials is presented. E, (i) Tube formation in MEEC ENG+/+ cells and endoglin null ENG−/− MEECs cells in the absence or presence of inhibin A after 16 hours. Graph of average number of meshes as described in Materials and Methods; scale bar, 500 μm. (ii) Western blotting for SMAD1/5 phosphorylation in response to 20 pmol/L inhibin A for up to 60 minutes in MEEC ENG+/+ or ENG−/− cells as indicated.

Close modal

SMAD1/5 activation positively regulates angiogenesis (31, 32) and our data indicate inhibin specifically induces SMAD1/5 activation in endothelial cells. These data suggest a requirement of endothelial-specific receptors for the observed inhibin response. Endoglin/CD105 is a well-established endothelial-specific TGFβ coreceptor that we, and several others, have shown to have key roles in transducing angiogenic signals (33). To test whether inhibin induced tube formation and SMAD1/5 phosphorylation in endothelial cells are dependent on endoglin, we employed three approaches, (i) a humanized monoclonal antibody to endoglin TRC105 that we and others have shown to neutralize endoglin function (34, 35), (ii) a previously characterized endoglin knockout mouse embryonic endothelial cell line (MEECs, ENG−/− and corresponding wild-type MEECs; refs. 15, 36), and (iii) endoglin shRNA. We find that inhibin-induced tube formation was significantly reduced by TRC105 compared with cells treated with inhibin alone (Fig. 4C). Both endoglin shRNA in HMEC-1s and endoglin knockout in MEEC ENG−/− abrogated inhibin-induced tube formation and SMAD1/5 activation (Fig. 4D and E, respectively). These data demonstrate that endoglin is a key mediator of inhibin responses.

Inhibin-induced signaling and tube formation require ALK1 and promote endoglin–ALK1 interactions in endothelial cells

SMAD1/5 is phosphorylated by specific serine threonine kinase type I receptors (37). We thus sought to identify the type I TGFβ receptor in inhibin-induced SMAD1/5 signaling and angiogenesis. We first determined the mRNA levels of different TGFβ type I receptors (ALK1-7) in HMEC-1′s as compared with epithelial SKOV3 cells, because inhibin-induced signaling was specific to endothelial cells (Supplementary Fig. S6, ii). qRT-PCR analysis (Supplementary Fig. S7, i) indicated higher ALK1 and ALK3 mRNA levels compared with ALK2, 4, 5, 6,7 in endothelial cells with limited ALK1 expression in SKOV3 cells (Supplementary Fig. S7, i, right graph). We thus used the ALK1/2 inhibitor ML347, to block ALK1 and partially block ALK2/3 (IC50 for ALK1,2,3 are 46 nmol/L, 32 nmol/L and 10 μmol/L, respectively). SB431542 and Dorsomorphin were used to block ALK5, 4,7 and ALK2, 3,6, respectively (38). We find that ML347 was able to suppress inhibin induced tube formation (Fig. 5A, 1.2 fold, ***, P < 0.001), and completely blocked inhibin induced SMAD1/5 activation (Fig. 5B). SB431542 and dorsomorphin had no significant impact on inhibin induced tube formation (Supplementary Fig. S7, ii). These data implicate ALK1 kinase activity in inhibin induced tube formation and signaling responses. To confirm a role for ALK1, shRNA to ALK1 (shALK1) was used in HMEC-1′s and compared with control HMEC-1′s. shALK1 cells were unable to mount an inhibin-mediated tubulogenesis response (Fig. 5C). shAlk1 HMEC-1 cells were also unable to increase SMAD1/5 phosphorylation in response to inhibin, to the same extent as shControl HMEC-1 cells (Fig. 5D). These data implicate ALK1 in mediating inhibin signaling in endothelial cells.

Figure 5.

Inhibin A induced tube formation and SMAD1/5 phosphorylation is mediated via ALK1 receptor kinases. Tube formation (A) and SMAD1/5 phosphorylation (B) in HMEC-1 cells in response to inhibin A in the absence and presence of 5 μmol/L of ML347 (ALK1,2 inhibitor) added 30 minutes before treatment with inhibin A. Graph of average number of meshes as described in Materials and Methods. C, qRT-PCR analysis of ALK1 mRNA in HMEC-1 cells transfected with either control (shControl) or ALK1 shRNA (shALK1) for 48 hours before tube formation assays in the absence or presence of recombinant inhibin A. D, Western blotting for SMAD1/5 phosphorylation in response to 20 pmol/L inhibin A treatments for up to 60 minutes. Graph of average number of meshes as described in Materials and Methods. Mean ± SEM; ***, P < 0.001 and **, P < 0.01, Student t test; scale bar, 500 μm. E, Endoglin-HA and ALK1-Myc receptor complex formation at the cell surface of COS7 cells either untreated or stimulated with 100 pmol/L inhibin A, followed by immunolabeling with anti-Myc and anti-HA antibodies at 4°C to capture cell surface interactions (Materials and Methods). The unprocessed images are raw images. Processed images represent a 20% reduction in intensity and colocalized images show only those pixels where red (ALK1-Myc) and green (ENG-HA) punctum overlap. Bar graph shows percentage of co-patched puncta of ALK1-Myc/Endoglin HA complexes. ***, P < 0.001 (n = 5 fields/condition), ANOVA with Holm-Sidak post hoc test; scale bar, 500 μm. F, Endogenous endoglin interactions with Myc-tagged Alk1 determined using a proximity ligation assay in MEEC ENG+/+ and MEEC ENG−/− either untreated or upon inhibin A treatment for 30 minutes; scale bar, 20 μm. Bar graph shows ALK1–ENG interactions normalized to nuclei (n = 10/condition).

Figure 5.

Inhibin A induced tube formation and SMAD1/5 phosphorylation is mediated via ALK1 receptor kinases. Tube formation (A) and SMAD1/5 phosphorylation (B) in HMEC-1 cells in response to inhibin A in the absence and presence of 5 μmol/L of ML347 (ALK1,2 inhibitor) added 30 minutes before treatment with inhibin A. Graph of average number of meshes as described in Materials and Methods. C, qRT-PCR analysis of ALK1 mRNA in HMEC-1 cells transfected with either control (shControl) or ALK1 shRNA (shALK1) for 48 hours before tube formation assays in the absence or presence of recombinant inhibin A. D, Western blotting for SMAD1/5 phosphorylation in response to 20 pmol/L inhibin A treatments for up to 60 minutes. Graph of average number of meshes as described in Materials and Methods. Mean ± SEM; ***, P < 0.001 and **, P < 0.01, Student t test; scale bar, 500 μm. E, Endoglin-HA and ALK1-Myc receptor complex formation at the cell surface of COS7 cells either untreated or stimulated with 100 pmol/L inhibin A, followed by immunolabeling with anti-Myc and anti-HA antibodies at 4°C to capture cell surface interactions (Materials and Methods). The unprocessed images are raw images. Processed images represent a 20% reduction in intensity and colocalized images show only those pixels where red (ALK1-Myc) and green (ENG-HA) punctum overlap. Bar graph shows percentage of co-patched puncta of ALK1-Myc/Endoglin HA complexes. ***, P < 0.001 (n = 5 fields/condition), ANOVA with Holm-Sidak post hoc test; scale bar, 500 μm. F, Endogenous endoglin interactions with Myc-tagged Alk1 determined using a proximity ligation assay in MEEC ENG+/+ and MEEC ENG−/− either untreated or upon inhibin A treatment for 30 minutes; scale bar, 20 μm. Bar graph shows ALK1–ENG interactions normalized to nuclei (n = 10/condition).

Close modal

ALK1, a primarily endothelial-specific receptor (39), forms stable complexes with endoglin, and this interaction is required for SMAD1/5 signaling (40). To test whether stable, cell surface ALK1–endoglin interactions could be regulated by inhibin, we used a live cell co-patching method (41). We transiently expressed HA-tagged endoglin (ENG-HA) and Myc-tagged ALK1 (ALK1-Myc) in COS7 cells and determined the change in percentage of co-patched receptors at the cell surface upon inhibin treatment. We find a 20% baseline interaction between ALK1-Myc and endoglin-HA, which increased significantly to 60% upon inhibin treatment (Fig. 5E, ***, P < 0.001), suggesting inhibin-dependent increases in ALK1-endoglin complexes at the cell surface. To complement the COS7 cell surface co-patch studies and confirm the effect of inhibin in the endothelial environment, we used a proximity ligation assay (42) in MEECs. We examined and quantified the proximity of ALK1-HA and endogenous endoglin in unstimulated and inhibin stimulated MEECs and find a significant increase in the ALK1 proximity interactions with endogenous endoglin in response to a 30 minutes inhibin treatment. ENG−/− MEECs were used as controls (Fig. 5F, ***, P < 0.001). Together, these findings strongly suggest a novel endoglin-ALK1 response to inhibin, a TGFβ family ligand with previously unknown functions in cancer and in endothelial cells. Taken together, we demonstrate a novel signaling ligand for angiogenesis in cancer.

Using a combination of antibody and shRNA-based approaches coupled with in vivo experimental models; we have identified and defined a novel paracrine regulator of tumor angiogenesis. Although inhibin has long been known to be elevated in multiple cancers no specific mechanism of action and outcome of elevated inhibin has been previously established. Our findings provide the first evidence of inhibin's ability to transduce signals and act as a pacracrine tumor angiogenic factor. Likewise, we provide the first use of an anti-inhibin α antibody to suppress angiogenesis. In accordance with prior studies, we observed inhibin α overexpression in a spectrum of ovarian cancers (Fig. 1) and observed increased inhibin α levels in tumors correlating with MVD in patients and xenograft tumors. Increased inhibin α also correlated with poor overall patient survival, particularly in high-grade serous ovarian cancer patients with alterations in TP53 and in renal clear cell carcinomas that are highly angiogenic. Notably, we demonstrate a novel function for inhibin that is dependent on endothelial specific TGFβ receptors endoglin and ALK1, which are upregulated during tumor angiogenesis and explored as angiogenic targets (43). Surprisingly, suppressing inhibin α expression and inhibin secretion in tumor cells had marginal effects on epithelial (autocrine) cell growth (Supplementary Fig. S4iv). However, in the intraperitoneal model that mimics human disease leading to ascites accumulation and peritoneal spread (44, 30), inhibin α shRNA significantly reduced ascites accumulation and peritoneal tumor burden in vivo (Fig. 3C), indicating not just a significant role for inhibin in metastasis but also the potential significance of inhibin induced angiogenesis in this process. Although our findings using inhibin α antibodies (Fig. 2) and recombinant inhibin in in vitro (Figs. 2, 4, 5) and in vivo Matrigel plug assays (Fig. 3A) indicate a direct role for inhibin on angiogenesis that is reduced upon INHA shRNA, one alternate mechanism for reduced metastasis observed using shINHA tumor cells could be alterations in pro-angiogenic factors secreted from the tumor cells that warrants future investigation.

Previous studies on inhibin signaling, emphasize the non-signaling interactions with the TGFβ coreceptor betaglycan via the α subunit in epithelial cells, highlighting a competitive antagonist model for inhibin with TGFβ members (13, 45). However, some outcomes of inhibin, including effects of inhibin in prostate cancer metastasis (11) and systems where inhibin does not antagonize activin (46), appear inconsistent with this model as inhibin's only mode action. Supporting the possibility of additional modes of action for inhibin, we find that activin primarily activates SMAD2/3 in endothelial cells (Supplementary Fig. S8), contrasting with inhibin, which primarily activates SMAD1/5 exclusively in endothelial cells (Fig. 4B; Supplementary Fig. S6ii). Because our findings demonstrate the critical role for endoglin (Fig. 4) in inhibin induced angiogenesis and signaling (Fig. 4), it is likely that the established model of inhibin's functional antagonism using betaglycan, may not be the same in endothelial cells that express endoglin.

In addition to inhibin, the ligands that use these receptors include TGFβ1, and BMP9/GDF2 that can activate both ALK1/SMAD1/5/8 and ALK5/Smad2/3 (1). We find that Activin significantly activates Smad2/3 with limited activation of SMAD1/5 (Supplementary Fig. S8). The SMAD2/3 pathway is an established antiangiogenic mechanism contrasting with SMAD1/5 as a proangiogenic mechanism (1). Consistently, activin did not support angiogenesis in vitro (Fig. 2) and was reported to suppress angiogenesis in vivo (47). Combining inhibin with other TGFβ members in vitro (Fig. 2), reveal that inhibin promotes angiogenesis to the same extent as BMP9 and can override both TGFβ and activin in inducing angiogenesis. It is likely that the specificity of SMAD1/5 activation in response to inhibin (unlike TGFβ, BMP9) represents an important distinction between inhibin and other TGFβ members in angiogenesis that necessitate future in depth analysis.

In summary, we provide new insights for inhibin in cancer, a hormone and TGFβ family member under investigation for over 90 years and provide a molecular and mechanistic basis for clinical outcomes observed in these cancers based on inhibin expression. Inhibin can have pronounced effects on angiogenesis and metastasis with broad therapeutic implications for other cancers as well. We thus propose inhibin as an attractive and safe target for ovarian cancer and other cancers prevalent in postmenopausal women where inhibin levels are normally very low, thereby limiting potential on target side effects. Further investigation into anti-inhibin therapeutic strategies, as a result, is warranted.

No potential conflicts of interest were disclosed.

Conception and design: P. Singh, K. Mythreye

Development of methodology: P. Singh, L.M. Jenkins, V. Alers, B. Győrffy, K. Mythreye

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.M. Jenkins, B. Horst, P. Kaur, K. Mythreye

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Singh, L.M. Jenkins, B. Horst, V. Alers, S. Pradhan, T. Srivastava, B. Győrffy, E.V. Broude, N.Y. Lee, K. Mythreye

Writing, review, and/or revision of the manuscript: P. Singh, L.M. Jenkins, B. Horst, N. Hempel, B. Győrffy, E.V. Broude, N.Y. Lee, K. Mythreye

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Singh, L.M. Jenkins, N.Y. Lee, K. Mythreye

Study supervision: E.V. Broude, K. Mythreye

We thank Drs. Shannon Davis and Ioluia Chatzistamou for help with mouse tissue processing and analysis of tissue cores, Andy Nixon and Miao at Duke University with optimizing inhibin ELISAs and the “University of South Carolina Center for Targeted Therapeutics Functional Genomics Core Facility” for lentiviral preparations and the “University of South Carolina Center for Targeted Therapeutics Microscopy and Flow Cytometry Core Facility” for help with imaging. This work was funded by the NIH grant 5P20-GM109091 and a grant from the Marsha Rivkin Foundation to Karthikeyan Mythreye, USC Aspire Fellowship to P. Singh and the USC SPARC Graduate Research Grant AAUW Dissertation Fellowship, and NIH Predoctoral F31 Fellowship, (GM122379-01; to L.M. Jenkins).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Pardali
E
,
ten Dijke
P
. 
Transforming growth factor-beta signaling and tumor angiogenesis
.
Front Biosci
2009
;
14
:
4848
61
.
2.
Overlie
I
,
Morkrid
L
,
Andersson
AM
,
Skakkebaek
NE
,
Moen
MH
,
Holte
A
. 
Inhibin A and B as markers of menopause: a five-year prospective longitudinal study of hormonal changes during the menopausal transition
.
Acta Obstet Gynecol Scand
2005
;
84
:
281
5
.
3.
Walentowicz
P
,
Krintus
M
,
Sadlecki
P
,
Grabiec
M
,
Mankowska-Cyl
A
,
Sokup
A
, et al
Serum inhibin A and inhibin B levels in epithelial ovarian cancer patients
.
PLoS One
2014
;
9
:
e90575
.
4.
Robertson
DM
,
Pruysers
E
,
Jobling
T
. 
Inhibin as a diagnostic marker for ovarian cancer
.
Cancer Lett
2007
;
249
:
14
7
.
5.
Walton
KL
,
Makanji
Y
,
Robertson
DM
,
Harrison
CA
. 
The synthesis and secretion of inhibins
.
Vitam Horm
2011
;
85
:
149
84
.
6.
Matzuk
MM
,
Finegold
MJ
,
Su
JG
,
Hsueh
AJ
,
Bradley
A
. 
Alpha-inhibin is a tumour-suppressor gene with gonadal specificity in mice
.
Nature
1992
;
360
:
313
9
.
7.
Shanbhag
SA
,
Sheth
AR
,
Nanivadekar
SA
,
Sheth
NA
. 
Immunoreactive inhibin-like material in serum and gastric juice of patients with benign and malignant diseases of the stomach
.
Br J Cancer
1985
;
51
:
877
82
.
8.
Suzuki
Y
,
Sugiyama
M
,
Abe
N
,
Fujioka
Y
,
Atomi
Y
. 
Immunohistochemical similarities between pancreatic mucinous cystic tumor and ovarian mucinous cystic tumor
.
Pancreas
2008
;
36
:
e40
6
.
9.
McCluggage
WG
,
Maxwell
P
. 
Adenocarcinomas of various sites may exhibit immunoreactivity with anti-inhibin antibodies
.
Histopathology
1999
;
35
:
216
20
.
10.
McCluggage
WG
,
Maxwell
P
,
Patterson
A
,
Sloan
JM
. 
Immunohistochemical staining of hepatocellular carcinoma with monoclonal antibody against inhibin
.
Histopathology
1997
;
30
:
518
22
.
11.
Balanathan
P
,
Williams
ED
,
Wang
H
,
Pedersen
JS
,
Horvath
LG
,
Achen
MG
, et al
Elevated level of inhibin-alpha subunit is pro-tumourigenic and pro-metastatic and associated with extracapsular spread in advanced prostate cancer
.
Br J Cancer
2009
;
100
:
1784
93
.
12.
Farnworth
PG
,
Harrison
CA
,
Leembruggen
P
,
Chan
KL
,
Stanton
PG
,
Ooi
GT
, et al
Inhibin binding sites and proteins in pituitary, gonadal, adrenal and bone cells
.
Mol Cell Endocrinol
2001
;
180
:
63
71
.
13.
Lewis
KA
,
Gray
PC
,
Blount
AL
,
MacConell
LA
,
Wiater
E
,
Bilezikjian
LM
, et al
Betaglycan binds inhibin and can mediate functional antagonism of activin signalling
.
Nature
2000
;
404
:
411
4
.
14.
Gatza
CE
,
Oh
SY
,
Blobe
GC
. 
Roles for the type III TGF-beta receptor in human cancer
.
Cell Signal
2010
;
22
:
1163
74
.
15.
Pece-Barbara
N
,
Vera
S
,
Kathirkamathamby
K
,
Liebner
S
,
Di Guglielmo
GM
,
Dejana
E
, et al
Endoglin null endothelial cells proliferate faster and are more responsive to transforming growth factor beta1 with higher affinity receptors and an activated Alk1 pathway
.
J Biol Chem
2005
;
280
:
27800
8
.
16.
Tian
H
,
Mythreye
K
,
Golzio
C
,
Katsanis
N
,
Blobe
GC
. 
Endoglin mediates fibronectin/alpha5beta1 integrin and TGF-beta pathway crosstalk in endothelial cells
.
EMBO J
2012
;
31
:
3885
900
.
17.
Gilboa
L
,
Wells
RG
,
Lodish
HF
,
Henis
YI
. 
Oligomeric structure of type I and type II transforming growth factor beta receptors: homodimers form in the ER and persist at the plasma membrane
.
J Cell Biol
1998
;
140
:
767
77
.
18.
Varadaraj
A
,
Jenkins
LM
,
Singh
P
,
Chanda
A
,
Snider
J
,
Lee
NY
, et al
TGF-beta triggers rapid fibrillogenesis via a novel TbetaRII dependent fibronectin trafficking mechanism
.
Mol Biol Cell
2017
;
28
:
1195
1207
.
19.
Malinda
KM
. 
In vivo matrigel migration and angiogenesis assay
.
Methods Mol Biol
2009
;
467
:
287
94
.
20.
DeCicco-Skinner
KL
,
Henry
GH
,
Cataisson
C
,
Tabib
T
,
Gwilliam
JC
,
Watson
NJ
, et al
Endothelial cell tube formation assay for the in vitro study of angiogenesis
.
J Vis Exp
2014
:
e51312
.
21.
Laib
AM
,
Bartol
A
,
Alajati
A
,
Korff
T
,
Weber
H
,
Augustin
HG
. 
Spheroid-based human endothelial cell microvessel formation in vivo
.
Nat Protoc
2009
;
4
:
1202
15
.
22.
Wu
SY
,
Rupaimoole
R
,
Shen
F
,
Pradeep
S
,
Pecot
CV
,
Ivan
C
, et al
A miR-192-EGR1-HOXB9 regulatory network controls the angiogenic switch in cancer
.
Nat Commun
2016
;
7
:
11169
.
23.
Stone
RL
,
Baggerly
KA
,
Armaiz-Pena
GN
,
Kang
Y
,
Sanguino
AM
,
Thanapprapasr
D
, et al
Focal adhesion kinase: an alternative focus for anti-angiogenesis therapy in ovarian cancer
.
Cancer Biol Ther
2014
;
15
:
919
29
.
24.
David
L
,
Feige
JJ
,
Bailly
S
. 
Emerging role of bone morphogenetic proteins in angiogenesis
.
Cytokine Growth Factor Rev
2009
;
20
:
203
12
.
25.
Jin
Y
,
Kaluza
D
,
Jakobsson
L
. 
VEGF, Notch and TGFbeta/BMPs in regulation of sprouting angiogenesis and vascular patterning
.
Biochem Soc Trans
2014
;
42
:
1576
83
.
26.
Pecot
CV
,
Rupaimoole
R
,
Yang
D
,
Akbani
R
,
Ivan
C
,
Lu
C
, et al
Tumour angiogenesis regulation by the miR-200 family
.
Nat Commun
2013
;
4
:
2427
.
27.
Tan
DS
,
Agarwal
R
,
Kaye
SB
. 
Mechanisms of transcoelomic metastasis in ovarian cancer
.
Lancet Oncol
2006
;
7
:
925
34
.
28.
Ahmed
N
,
Stenvers
KL
. 
Getting to know ovarian cancer ascites: opportunities for targeted therapy-based translational research
.
Front Oncol
2013
;
3
:
256
.
29.
Shaw
TJ
,
Senterman
MK
,
Dawson
K
,
Crane
CA
,
Vanderhyden
BC
. 
Characterization of intraperitoneal, orthotopic, and metastatic xenograft models of human ovarian cancer
.
Mol Ther
2004
;
10
:
1032
42
.
30.
Connolly
DC
,
Hensley
HH
. 
Xenograft and transgenic mouse models of epithelial ovarian cancer and non-invasive imaging modalities to monitor ovarian tumor growth in situ: applications in evaluating novel therapeutic agents
.
Curr Protoc Pharmacol
2009
;45:14.12.1–14.12.26.
31.
Goumans
MJ
,
Valdimarsdottir
G
,
Itoh
S
,
Lebrin
F
,
Larsson
J
,
Mummery
C
, et al
Activin receptor-like kinase (ALK)1 is an antagonistic mediator of lateral TGFbeta/ALK5 signaling
.
Mol Cell
2003
;
12
:
817
28
.
32.
David
L
,
Mallet
C
,
Mazerbourg
S
,
Feige
JJ
,
Bailly
S
. 
Identification of BMP9 and BMP10 as functional activators of the orphan activin receptor-like kinase 1 (ALK1) in endothelial cells
.
Blood
2007
;
109
:
1953
61
.
33.
Nassiri
F
,
Cusimano
MD
,
Scheithauer
BW
,
Rotondo
F
,
Fazio
A
,
Yousef
GM
, et al
Endoglin (CD105): a review of its role in angiogenesis and tumor diagnosis, progression and therapy
.
Anticancer Res
2011
;
31
:
2283
90
.
34.
Rosen
LS
,
Hurwitz
HI
,
Wong
MK
,
Goldman
J
,
Mendelson
DS
,
Figg
WD
, et al
A Phase 1 First-in-Human Study of TRC105 (Anti-Endoglin Antibody) in Patients with Advanced Cancer
.
Clin Cancer Res
2012
;
18
:
4820
9
.
35.
Kumar
S
,
Pan
CC
,
Bloodworth
JC
,
Nixon
AB
,
Theuer
C
,
Hoyt
DG
, et al
Antibody-directed coupling of endoglin and MMP-14 is a key mechanism for endoglin shedding and deregulation of TGF-beta signaling
.
Oncogene
2014
;
33
:
3970
9
.
36.
Barbara
NP
,
Wrana
JL
,
Letarte
M
. 
Endoglin is an accessory protein that interacts with the signaling receptor complex of multiple members of the transforming growth factor-beta superfamily
.
J Biol Chem
1999
;
274
:
584
94
.
37.
Massague
J
,
Attisano
L
,
Wrana
JL
. 
The TGF-beta family and its composite receptors
.
Trends Cell Biol
1994
;
4
:
172
8
.
38.
Inman
GJ
,
Nicolas
FJ
,
Callahan
JF
,
Harling
JD
,
Gaster
LM
,
Reith
AD
, et al
SB-431542 is a potent and specific inhibitor of transforming growth factor-beta superfamily type I activin receptor-like kinase (ALK) receptors ALK4, ALK5, and ALK7
.
Mol Pharmacol
2002
;
62
:
65
74
.
39.
Bertolino
P
,
Deckers
M
,
Lebrin
F
,
ten Dijke
P
. 
Transforming growth factor-beta signal transduction in angiogenesis and vascular disorders
.
Chest
2005
;
128
:
585S
90S
.
40.
Pomeraniec
L
,
Hector-Greene
M
,
Ehrlich
M
,
Blobe
GC
,
Henis
YI
. 
Regulation of TGF-beta receptor hetero-oligomerization and signaling by endoglin
.
Mol Biol Cell
2015
;
26
:
3117
27
.
41.
Gilboa
L
,
Nohe
A
,
Geissendorfer
T
,
Sebald
W
,
Henis
YI
,
Knaus
P
. 
Bone morphogenetic protein receptor complexes on the surface of live cells: a new oligomerization mode for serine/threonine kinase receptors
.
Mol Biol Cell
2000
;
11
:
1023
35
.
42.
Soderberg
O
,
Gullberg
M
,
Jarvius
M
,
Ridderstrale
K
,
Leuchowius
KJ
,
Jarvius
J
, et al
Direct observation of individual endogenous protein complexes in situ by proximity ligation
.
Nat Methods
2006
;
3
:
995
1000
.
43.
Jonker
L
. 
TGF-beta & BMP receptors endoglin and ALK1: overview of their functional role and status as antiangiogenic targets
.
Microcirculation
2014
;
21
:
93
103
.
44.
Jazaeri
AA
,
Bryant
JL
,
Park
H
,
Li
H
,
Dahiya
N
,
Stoler
MH
, et al
Molecular requirements for transformation of fallopian tube epithelial cells into serous carcinoma
.
Neoplasia
2011
;
13
:
899
911
.
45.
Looyenga
BD
,
Wiater
E
,
Vale
W
,
Hammer
GD
. 
Inhibin-A antagonizes TGFbeta2 signaling by down-regulating cell surface expression of the TGFbeta coreceptor betaglycan
.
Mol Endocrinol
2010
;
24
:
608
20
.
46.
Wang
QF
,
Tilly
KI
,
Tilly
JL
,
Preffer
F
,
Schneyer
AL
,
Crowley
WF
 Jr
, et al
Activin inhibits basal and androgen-stimulated proliferation and induces apoptosis in the human prostatic cancer cell line, LNCaP
.
Endocrinology
1996
;
137
:
5476
83
.
47.
Kaneda
H
,
Arao
T
,
Matsumoto
K
,
De Velasco
MA
,
Tamura
D
,
Aomatsu
K
, et al
Activin A inhibits vascular endothelial cell growth and suppresses tumour angiogenesis in gastric cancer
.
British journal of cancer
2011
;
105
:
1210
7
.

Supplementary data