Autocrine activin A signalling in ovarian cancer cells regulates secretion of interleukin 6, autophagy, and cachexia

Abstract

Background: The majority of patients with advanced cancer develop cachexia, a weight loss syndrome that severely reduces quality of life and limits survival. Our understanding of the underlying mechanisms that cause the condition is limited, and there are currently no treatment options that can completely reverse cachexia. Several tumour-derived factors and inflammatory mediators have been suggested to contribute to weight loss in cachectic patients. However, inconsistencies between studies are recurrent. Activin A and interleukin 6 (IL-6) are among the best studied factors that seem to be important, and several studies support their individual role in cachexia development.

Methods: We investigated the interplay between activin A and IL-6 in the cachexia-inducing TOV21G cell line, both in culture and in tumours in mice. We previously found that the human TOV21G cells secrete IL-6 that induces autophagy in reporter cells and cachexia in mice. Using this established cachexia cell model, we targeted autocrine activin A by genetic, chemical, and biological approaches. The secretion of IL-6 from the cancer cells was determined in both culture and tumour-bearing mice by a species-specific ELISA. Autophagy reporter cells were used to monitor the culture medium for autophagy-inducing activities, and muscle mass changes were evaluated in tumour-bearing mice.

Results: We show that activin A acts in an autocrine manner to promote the synthesis and secretion of IL-6 from cancer cells. By inhibiting activin A signalling, the production of IL-6 from the cancer cells is reduced by 40-50% (up to 42% reduction on protein level, P = 0.0048, and 48% reduction on mRNA level, P = 0.0308). Significantly reduced IL-6 secretion (P < 0.05) from the cancer cells is consistently observed when using biological, chemical, and genetic approaches to interfere with the autocrine activin A loop. Inhibiting activin signalling also reduces the ability of the cancer cells to accelerate autophagy in non-cancerous cells (up to 43% reduced autophagy flux, P = 0.0006). Coherent to the in vitro data, the use of an anti-activin receptor 2 antibody in cachectic tumour-bearing mice reduces serum levels of cancer cell-derived IL-6 by 62% (from 417 to 159 pg/mL, P = 0.03), and, importantly, it reverses cachexia and counteracts loss of all measured muscle groups (P < 0.0005).

Conclusions: Our data support a functional link between activin A and IL-6 signalling pathways and indicate that interference with activin A-induced IL-6 secretion from the tumour has therapeutic potential for cancer-induced cachexia.

Keywords: Activin; Autocrine loop; Autophagy; Cachexia; IL-6.

Figure 1

Activin A acts in an autocrine or paracrine loop to promote the secretion of autophagy‐accelerating compounds from TOV21G cells. (A) Lower panel: Autophagy flux in autophagy reporter cells treated with recombinant activin A for 3 days at indicated concentrations. Mean from three independent experiments, each using triplicate wells ±SD. Upper panel: Level of phosphorylated SMAD3 (Ser423/Ser425) in autophagy reporter cells following 5 min exposure to activin A (50 ng/mL). β‐Tubulin is used as a loading control. (B) INHBA, ACVR2A, and ACVR2B transcript levels in TOV21G cells treated with siRNA targeting INHBA, ACVR2A, and ACVR2B, respectively, 1 day post‐transfection, relative to non‐targeting (NT) siRNA‐treated TOV21G cells. Measured using qPCR. Mean from six independent experiments ±SD. ^*** P < 0.0005 vs. NT siRNA (Student's t‐test). (C) Level of activin A protein in CM from TOV21G cells treated with INHBA siRNA, 3 days post‐transfection. Mean from two experiments. (D) Autophagy flux in autophagy reporter cells treated with CM from TOV21G cells. CM had been collected 1 or 3 days post‐transfection using NT siRNA or siRNA targeting INHBA, ACVR2A, or ACVR2B. Mean from six independent experiments, each using triplicate wells. ^* P < 0.05, ^** P < 0.005 vs. respective NT siRNA (Student's t‐test).

Figure 2

KRAS affects the secretion of the autophagy‐inducing cytokine IL‐6 from TOV21G cells. (A) KRAS transcript level in TOV21G cells treated with two different siRNAs targeting KRAS, 1 day post‐transfection, relative to non‐targeting (NT) siRNA‐treated TOV21G cells. Measured using qPCR. Mean from three independent experiments ±SD. ^*** P < 0.0005 and ^** P < 0.005 vs. NT siRNA (Student's t‐test). (B) IL‐6 transcript level in TOV21G cells treated with siRNAs targeting KRAS, 1 day post‐transfection, relative to NT siRNA‐treated TOV21G cells. Measured using qPCR. Mean from three independent experiments ±SD. ^* P < 0.05 vs. NT siRNA (Student's t‐test). (C) Relative level of IL‐6, measured using an ELISA assay, in CM from TOV21G cells treated with NT siRNA or siRNAs targeting KRAS. CM was harvested 1 day post‐transfection. Mean from three experiments ±SD. ^* P < 0.05 vs. NT siRNA (Student's t‐test).

Figure 3

Activin A affects the secretion of the autophagy‐inducing cytokine IL‐6 from TOV21G cells. (A) Relative level of IL‐6, measured using an ELISA assay, in CM from TOV21G cells treated with non‐targeting (NT) siRNA or siRNA targeting INHBA, ACVR2A, or ACVR2B. CM was harvested 1 or 3 days post‐transfection. Mean from three experiments [ACVR2A (3 days) and ACVR2B (1 day and 3 days)], four experiments [ACVR2A (1 day) and INHBA (3 days)], or six experiments [INHBA (1 day)] ±SD. ^* P < 0.05, ^** P < 0.005, ^*** P < 0.0005 vs. respective NT siRNA (Student's t‐test). (B) Relative level of IL‐6, measured using an ELISA assay, in CM from TOV21G cells treated with 1.6 μg/mL activin neutralizing antibody, 1.8 μg/mL ActRIIB decoy receptor, or 10 μM ALK4/5/7 inhibitor (SB431542). CM was harvested after 1 day of treatment. Mean from three experiments ±SD. ^* P < 0.05 vs. control (Student's t‐test). (C) SMAD3 transcript level in TOV21G cells treated with siRNA targeting SMAD3, 1 day post‐transfection, relative to NT siRNA‐treated TOV21G cells. Measured using qPCR. Mean from three independent experiments ±SD. ^*** P < 0.0005 vs. NT siRNA (Student's t‐test). (D) Relative level of IL‐6, measured using an ELISA assay, in CM from TOV21G cells treated with NT siRNA or siRNA targeting SMAD3. CM was harvested 1 day post‐transfection. Mean from six experiments ±SD. ^*** P < 0.0005 vs. NT siRNA (Student's t‐test). (E) IL‐6 transcript level in TOV21G cells treated with NT siRNA (±50 ng/mL recombinant activin A) or siRNA targeting INHBA (±50 ng/mL recombinant activin A), ACVR2A, or ACVR2B. Measured using qPCR. RNA was isolated 1 or 3 days post‐transfection. Mean from 10 independent experiments (seven for recombinant activin A) ±SD. ^* P < 0.05, ^** P < 0.005 (Student's t‐test). (F) IL‐6 transcript level in TOV21G cells treated with 10 μM ALK4/5/7 inhibitor (SB431542) for 1 day, relative to vehicle‐treated TOV21G cells. Measured using qPCR. Mean from three independent experiments ±SD. ^* P < 0.05 vs. vehicle (Student's t‐test). (G) IL‐6 transcript level in TOV21G cells treated with siRNA targeting SMAD3, 1 day post‐transfection, relative to NT siRNA‐treated TOV21G cells. Measured using qPCR. Mean from three independent experiments ±SD. ^* P < 0.05 vs. NT siRNA (Student's t‐test). (H, I) Levels of pNF‐κB p65 (Ser536) and Erk1/2, and p‐p38 (Thr180/Tyr182) and Erk1/2, respectively, in TOV21G cells treated with NT siRNA or siRNA targeting INHBA, ACVR2A, or ACVR2B. Extracts were made 3 days post‐transfection. Representative blots are shown with numbers representing mean values from five [panel (H)] or four [panel (I)] independent experiments (and SD) after normalization against signals from Erk1/2. ^* P < 0.05, ^** P < 0.005, ^*** P < 0.0005 (Student's t‐test).

Figure 4

Interference with activin A signalling reduces tumour‐derived IL‐6 in sera and reverses cachexia in mice. (A, B) Mean relative body weight (at indicated time intervals) and mean relative change in body weight ±SEM, respectively, of non‐tumour control mice (n = 5) and vehicle‐treated (n = 7) and CDD866‐treated (n = 7) TOV21G tumour‐bearing mice. For the vehicle group, the data on Days 30, 32, and 35 are based on six, five, and four mice, respectively. (C–F) Mean relative change in tibialis, gastrocnemius complex, quadriceps, and heart weight, respectively, ±SEM of vehicle‐treated (n = 7) and CDD866‐treated (n = 7) TOV21G tumour‐bearing mice relative to non‐tumour control mice (n = 5). (G, H) Mean tumour volume and tumour weight ±SEM, respectively, of vehicle‐treated (n = 7) and CDD866‐treated (n = 7) TOV21G tumour‐bearing mice. (I) Mean relative change in white adipose tissue weight ±SEM of vehicle‐treated (n = 7) and CDD866‐treated (n = 7) TOV21G tumour‐bearing mice relative to non‐tumour control mice (n = 5). (J–L) Activin A, human IL‐6 (hIL‐6), and murine IL‐6 (mIL‐6) protein level, respectively, in sera from non‐tumour control mice (n = 5) and vehicle‐treated (n = 7) and CDD866‐treated (n = 7) TOV21G tumour‐bearing mice. Mean ± SEM are indicated. ^* P < 0.05, ^** P < 0.005, ^*** P < 0.0005 (Student's t‐test), n.s. = non‐significant.

Figure 5

Association between INHBA and IL‐6 gene expression in human ovarian tumours. IL‐6 and INHBA mRNA expression in human ovarian serous cystadenocarcinoma (n = 307). From the cBioPortal database (http://www.cbioportal.org/).