BM-hMSC secretome elicits anti-proliferative and apoptotic effects in H295R cells
H295R cells were incubated with BM-hMSC secretome to evaluate therapeutic potential. After 24 hours, we observed a significant reduction (7.96% ± 0.23) in cell growth rate, as measured by Ki-67 protein expression compared with control media-treated cells (12.37% ± 0.19; Fig. 1a). Additionally, secretome treatment significantly increased both early apoptosis (74.38% ± 1.00; Fig. 1b) as well as late apoptosis and necrosis (2.43% ± 0.21; Fig. 1c), as measured by Annexin-V and Annexin-V/7-AAD expression, respectively, compared with the control group (64.94% ± 1.47 and 1.30% ± 0.43). Thus, our results indicate that BM-hMSC secretome inhibits growth of H295R cells.
BM-hMSC secretome decreases steroidogenesis-related gene expression and androgen production in H295R cells
We previously reported that CYP17A1, CYP11A1, and DENND1A, key genes for ovarian androgen biosynthesis, are upregulated in PCOS-theca cells compared with healthy theca cells (27, 33). Hence, we evaluated the effect of the BM-hMSC secretome on the expression of these genes using our in vitro model. Secretome treatment resulted in significant downregulation of CYP17A1 (0.56 ± 0.02 fold) and DENND1A (0.37 ± 0.05 fold) gene expression in H295R cells compared with media-treated cells (Fig. 1d-f). However there was no significant decrease in CYP11A1 gene expression (0.82 ± 0.05 fold, p=0.127). We confirmed these findings at the protein level using immunoblot analysis, which showed that CYP17A1 (0.84 ± 0.02 fold) and DENND1A (0.26 ± 0.01 fold) were significantly decreased in secretome-treated H295R cells compared with the control group, while no change was observed in CYP11A1 (0.97 ± 0.02 fold, p=0.29; Fig. 1g-i). We validated these observations in PCOS patient-derived theca cells (n=2) treated with BM-hMSC secretome. Secretome treatment significantly downregulated CYP17A1 (Patient 1: 0.36 ± 0.20 fold, Patient 2: 0.05 ± 0.04 fold) gene expression (Fig. 1j) and protein expression (0.40 ± 0.39 fold; Fig. 1k) in theca cells from both patients compared with media-treated controls.
We investigated the effects of steroidogenesis-related gene inhibition by the BM-hMSC secretome on testosterone secretion.We explored whether steroidogenesis-related gene inhibition by the BM-hMSC secretome affected testosterone secretion. Compared with the media control group (474.6±27.5 ng/dL), secretome treatment suppressed testosterone secretion in H295R cells (267.7±4.0 ng/dL) (Fig. 1l). Testosterone secretion was also suppressed in human PCOS theca cells (146.4±13.4 ng/dL) (Fig. 1m) compared with the media control group (214.7±11.8 ng/dL). In summary, our data indicate that BM-hMSC secreted factors inhibit androgen production.
BM-hMSC secretome exerts an anti-inflammatory effect on H295R cells
Chronic inflammation is a major factor affecting the ovarian microenvironment in patients with PCOS inducing higher ovarian androgen production (34-37), that involves two pro-inflammatory cytokines, interleukin-1 beta (IL-1β) and tumor necrosis factor (TNF-α) (38). Treatment of H295R cells with BM-hMSC secretome significantly downregulated gene expression of IL-1β (IL1B: 0.04 ± 0.003 fold) and TNF-α (TNFA: 0.85 ± 0.02 fold) compared with the control media group (Fig. 1n, o), indicating a decreased inflammatory response after treatment.
IL-10 decreases steroidogenesis-related gene expression and androgen production in H295R cells
Based on the observed anti-inflammatory effect of the BM-hMSC secretome, we tested the effect of the anti-inflammatory cytokine, IL-10, which is known to be released by MSC (31, 39, 40). IL-10 exerts immune-suppressive and anti-inflammatory effects in several disorders, including PCOS (31, 41). Recent reports highlighted a significantly lower serum level of IL-10 in women with PCOS compared with age- and BMI-matched healthy controls (42). High IL-10 levels may also increase insulin sensitivity by ameliorating the inflammatory responses to TNF-α and IL-6, which contribute to insulin resistance in PCOS (43, 44). First, we explored IL-10 secretion from BM-MSCs by ELISA of conditioned media. We found a high concentration of IL-10 (164.8 pg/ml) compared with control media (1.32 pg/ml; Fig. 2a). We then examined the effect of IL-10 on steroidogenesis-related gene expression and androgen production in H295R cells. As shown in Figure 2, recombinant human IL-10 treatment significantly downregulated the expression of CYP17A1 (Control: 1.03 ± 0.01 fold, 125 pg/ml: 0.91 ± 0.01, 250 pg/ml: 0.89 ± 0.00 fold, 500 pg/ml: 0.86 ± 0.04 fold) and CYP11A1 (Control: 1.00 ± 0.03 fold, 125 pg/ml: 0.84 ± 0.02, 250 pg/ml: 0.87 ± 0.01 fold, 500 pg/ml: 0.86 ± 0.00 fold), and DENND1A (Control: 1.00 ± 0.02 fold, 125 pg/ml: 0.87 ± 0.01, 250 pg/ml: 0.92 ± 0.03 fold, 500 pg/ml: 0.86 ± 0.02 fold) in H295R cells compared with untreated controls (Fig. 2b-d). Additionally, IL-10 significantly decreased testosterone (Control: 1.56 ± 0.02 ng/ml, 125 pg/ml: 1.52 ± 0.01 ng/ml, 250 pg/ml: 1.42 ± 0.06 ng/ml, 500 pg/ml: 1.32 ± 0.10 ng/ml) and androstenedione (Control: 1.55 ± 0.03 ng/ml, 125 pg/ml: 1.44 ± 0.04 ng/ml, 250 pg/ml: 1.34 ± 0.04 ng/ml, 500 pg/ml: 1.38 ± 0.10 ng/ml) secretion from H295R cells in a dose-dependent manner compared with the control group (Fig. 2e, f). Our data suggest that IL-10 inhibits androgen production by regulating steroidogenic gene expression.
IL-10 exerts an anti-inflammatory effect on H295R cells
Next, we explored the anti-inflammatory effects of IL-10 on H295R cells by measuring expression of key pro-inflammatory cytokines, IL-6, TNF-α, and IL-1β, following IL-10 treatment. All tested concentrations of IL-10 significantly downregulated IL6 (Control: 1.00 ± 0.08 fold, 125 pg/ml: 0.73 ± 0.03 fold, 250 pg/ml: 0.79 ± 0.02 fold, 500 pg/ml: 0.70 ± 0.11 fold), TNFA (Control: 1.00 ± 0.04 fold, 125 pg/ml: 0.44 ± 0.02 fold, 250 pg/ml: 0.59 ± 0.07 fold, 500 pg/ml: 0.54 ± 0.04 fold), and IL1B (Control: 1.00 ± 0.04 fold, 125 pg/ml: 0.84 ± 0.02 fold, 250 pg/ml: 0.60 ± 0.08 fold, 500 pg/ml: 0.76 ± 0.1 fold) gene expression compared with untreated controls (Fig. 2g-i). Together, these data suggest that IL-10 is a key mediator of the effect of the BM-hMSC secretome on in vitro human cell PCOS models.
BM-hMSC reverse the metabolic phenotypes in an LTZ-induced PCOS mouse model
Next, we evaluated the potential therapeutic effects of BM-hMSC in vivo by injecting BM-hMSC into the ovaries of the well-established LTZ-induced PCOS mouse model (19). Five weeks after LTZ implantation, the mice in the PCOS group were significantly heavier (21.1 ± 0.25 grams) compared with age-matched control mice (19.3 ± 0.60 grams) that had received placebo pellets (Fig. 3a, b). Since PCOS women have insulin resistance and impaired glucose tolerance (45), we also performed a glucose tolerance test (GTT) and measured energy expenditure in PCOS mice before (5 weeks after LTZ or placebo) and 2 weeks after BM-hMSC engraftment (7 weeks after LTZ or placebo). Interestingly, PCOS mice treated with BM-hMSC exhibited a normal glucose tolerance profile compared with untreated PCOS mice (Fig. 3c-e). Moreover, we found that the untreated PCOS group had lower energy expenditure, based on a significant difference in thermogenesis, compared with PCOS mice treated with BM-hMSC (Fig. 3f-i).
The increase in thermogenesis in BM-hMSC-treated PCOS mice encouraged us to further evaluate fat metabolism in treated versus control PCOS mice. A previous study revealed the role of brown fat cells in the regulation of total energy expenditure (46). A process called “browning,” which refers to the transition of white fat into brown fat, is associated with upregulation of UCP-1 (47). Therefore, we stained white fat tissues collected from BM-hMSC-treated and untreated PCOS mice with UCP-1 and found greater proportions of brown-like fat cells, suggesting increased browning of white fat, in the BM-hMSC-treated group (Fig. 3j). At the molecular level, the browning process is regulated by several genes that control multiple aspects of mitochondrial activity, such as Pgc-1α, Cidea, and Prdm-16 (48). qPCR confirmed our UCP-1 immunohistochemistry results and showed significant upregulation of Ucp1 (21.00 ± 0.67 fold), Pgc1a (2.61 ± 0.08 fold), Cidea (3.78 ± 0.31 fold), and Prdm16 (5.15 ± 0.19 fold) gene expression in white fat collected from the BM-hMSC-treated PCOS mice compared with the untreated PCOS mice (Fig. 3k-n).
We also explored marker expression levels in brown fat tissue, and found that BM-hMSC treatment increases brown fat-related marker expression even in brown fat tissue (Fig S4). These results suggest that BM-hMSC can regulate adipose tissue metabolism by ameliorating inflammation and promoting brown fat formation.
BM-hMSC normalize the adipokine profile in adipose tissue in an LTZ-induced PCOS mouse model
Weight gain associated with LTZ-induced PCOS is partially due to white fat expansion (19). The expansion of gonadal fat in our LTZ-induced PCOS mice was marked by characteristic morphologic enlargement of fat cells detected by H&E staining (Fig S5a). Remarkably, the average size of adipocytes in the BM-hMSC-treated PCOS mice was significantly smaller than that in the untreated PCOS mice (Fig S5b), approaching the normal size range of adipocytes.
White fat adipocyte expansion is usually associated with an increase in leptin that correlates inversely with adiponectin levels (49). Studies have shown that adiponectin is a pivotal adipokine that can reverse PCOS metabolism (50), acting as a humoral factor that regulates fat homeostasis by establishing cross-talk between white and brown fat cells (51). To explore such cross-talk in our PCOS mouse model, we measured gene expression of leptin and adiponectin in brown adipose tissue as well as white gonadal fat using qPCR. Treatment with BM-hMSC upregulated adiponectin and downregulated leptin expression, thus normalizing the ratio of leptin to adiponectin in brown fat tissue compared with the untreated group (Fig S5c-e). Similar findings were observed in the white gonadal fat (Fig S5f-h), highlighting the ability of BM-hMSC to normalize fat metabolism in our PCOS mouse model.
Serum hormone analysis
To assess the endocrine status following BM-hMSC engraftment, total serum hormone levels in BM-hMSC-treated and untreated PCOS animals were measured. Serum T levels were significantly higher in the untreated PCOS group versus healthy controls, with no significant difference in the BM-hMSC-treated group (p=0.797). Furthermore, there were no changes in serum estrogen levels among the three groups. However, LH was significantly lower in the PCOS group than healthy controls, and LH levels decreased after BM-hMSC engraftment in PCOS mice. In addition, FSH levels were lower in the PCOS group compared with healthy controls and increased after BM-hMSC treatment, though the change was not statistically significant (Fig S6a-d).
BM-hMSC treatment reverses endometrial abnormalities in an LTZ-induced PCOS mouse model
PCOS imparts abnormalities in endometrial tissue, such as the thickening of endometrium epithelial cells and aberration of steroid receptor gene expression (52-54). Consequently, we analyzed endometrial tissue in BM-hMSC-treated versus untreated PCOS mice. The endometrial tissue of the PCOS group showed abnormal thickness (Fig S7a) and the AIB1 gene, known to be elevated in the PCOS endometrium (54), showed significant alterations in the PCOS group endometrium. These abnormalities were reversed in the BM-hMSC-treated PCOS group (Fig S7b). Similarly, steroid receptor genes AR and ERβ trended higher in the PCOS group and this was reversed after BM-hMSC treatment (Fig S7c, d).
Interestingly, the proliferation marker Ki-67 was significantly upregulated in BM-hMSC-treated PCOS mice compared with the untreated PCOS group (Fig S7e). Additionally, several inflammatory regulator genes such as IL6, IL16, CCL2, and TNF-α were higher in the PCOS group endometrium compared with normal control endometrium, and these changes were significantly reversed after BM-hMSC treatment (Fig S7f-i). These results suggest that intra-ovarian injection of BM-hMSC reversed various alterations in PCOS endometrium, at least partially, by normalizing steroid hormone receptors and inflammatory cytokine gene expression. These changes likely improved the quality of endometrium and contributed to the improved reproductive outcomes in the PCOS mice after BM-hMSC treatment.
BM-hMSC restore fertility in an LTZ-induced PCOS mouse model
To explore the effect of BM-hMSC treatment on reproductive function, we first analyzed ovarian morphology in BM-hMSC-treated versus untreated PCOS mice. Ovaries from untreated PCOS mice displayed typical PCOS characteristics, including lack of corpora lutea and antral follicles compared with untreated normal control mice (Fig. 4a). After intra-ovarian engraftment of BM-hMSC in both ovaries of PCOS mice, normal ovarian morphology was partially restored, including the reappearance of corpora lutea and antral follicles, as well as well-ordered stroma that was morphologically similar to that of the normal control group. These morphological changes suggest that BM-hMSC engraftment improved the pathological changes in PCOS ovaries and may potentially restore ovulation in PCOS mice (Fig. 4a).
Next, we performed a breeding experiment to test BM-hMSC’s treatment capacity to restore fertility in our PCOS mouse model. We found that healthy control mice had a higher rate of fertility (80%) than the subfertile PCOS group (10%). Interestingly, the pregnancy rate in BM-hMSC-treated PCOS mice was restored to a rate equal to that of the control group (Fig. 4b).
We also counted the number of delivered pups in all experimental groups. As shown in Figure 4d-g, we found that most PCOS mice were infertile. The number of pups delivered in the BM-hMSC-treated PCOS group was equivalent to the number delivered in the control group. Moreover, the average number of pups delivered per mouse in the BM-hMSC-treated PCOS group (5.5 ± 1.1) was significantly higher than that in the untreated PCOS group (0.8 ± 1.7; Fig. 4g). We also found no significant differences in the average body weight of the delivered pups between the control group and BM-hMSC-treated PCOS group at 0, 5, and 10 postnatal days (Fig. 4h). Notably, we did not observe any apparent morphological abnormalities in any pups during the study period. We further tested the effect of injecting the BM-hMSC secretome in our PCOS mouse model (Fig S8). Fertility of PCOS mice was restored in secretome-treated animals. These results suggest that either BM-hMSC or its secretome can restore impaired fertility in an LTZ-induced PCOS mouse model with no detectable abnormalities in the delivered newborns.
BM-hMSC restore normal ovarian gene expression in an LTZ-induced PCOS mouse model
PCOS abnormalities include enhanced androgen production and altered ovarian angiogenesis (55). We next examined the effect of BM-hMSC treatment on these abnormalities in our PCOS mice in vivo, to validate our in vitro data on the secretome effects on ovarian steroidogenic gene and inflammation marker expression. Mouse ovarian tissues from BM-hMSC-treated and untreated animals were analyzed for RNA and protein levels of steroidogenesis and angiogenesis markers. Cyp17a1 gene expression was significantly elevated in PCOS ovaries (13.73 ± 5.78 fold), which was significantly reversed after BM-hMSC treatment (1.22 ± 0.20 fold; Fig. 5a). Cyp19a1 (0.14 ± 0.02 fold) and Fshr (0.03 ± 0.01 fold) gene expression were lower in PCOS group ovaries, consistent with the prior characterization of the LTZ-induced PCOS mouse model (19); levels of both genes significantly increased after BM-hMSC treatment (Cyp19a1: 0.93 ± 0.11 fold, Fshr: 0.80 ± 0.09 fold; Fig. 5b, c). Previous studies reported an abnormal increase in ovarian angiogenesis in PCOS (55, 56); thus, we measured gene expression of angiogenesis marker Vegfa in untreated and BM-hMSC-treated PCOS mice. Gene expression of Vegfa was elevated in the PCOS group (7.50 ± 3.69 fold) and decreased after BM-hMSC treatment (0.48 ± 0.29 fold; Fig. 5d). Immunoblot analysis supported these results, where CYP17A1 was higher in the PCOS group (2.21 ± 0.14 fold) and significantly decreased after BM-hMSC treatment (1.37 ± 0.12 fold; Fig. 5e, f). Moreover, VEGF-A protein expression was elevated in PCOS mice (1.28 ± 0.15 fold) and decreased after BM-hMSC treatment (1.08 ± 0.38 fold), although the change was not statistically significant (Fig. 5e, g). Taken together, our in vivo results suggest that BM-hMSC treatment can inhibit androgen synthesis and angiogenesis consistent with the effect of the BM-hMSC secretome on H295R cells.
BM-hMSC secretome improves metabolic and reproductive phenotypes in LTZ-induced PCOS mice
Our in vitro and in vivo data suggest that the favorable effects of BM-hMSC engraftment likely occur in a paracrine fashion via secreted humoral factors in the BM-hMSC secretome. To explore the paracrine effect of BM-hMSC in the LTZ-induced PCOS mouse model, we delivered the BM-hMSC secretome by direct intra-ovarian injection into the ovaries of mice, and various metabolic and reproductive parameters were assessed. Analysis of white fat demonstrated a significant reduction in the size of fat cells in the secretome-treated PCOS group compared with the untreated PCOS group (Fig S8a), and the expression of UCP1 in brown fat tissue was also significantly higher in the secretome-treated PCOS group compared with untreated PCOS group (Fig S8b).
Morphological comparison of ovaries among the groups of mice by H&E staining revealed that secretome-treated PCOS ovaries had more antral follicles compared with untreated PCOS group ovaries (Fig S8c). Importantly, secretome treatment also restored fertility in PCOS mice compared with the untreated PCOS control group (Fig S8d, e). These results confirm the ability of the BM-hMSC secretome to reverse metabolic and reproductive abnormalities in the LTZ-induced PCOS mouse model and suggest that the positive effects of BM-hMSC are primarily mediated via paracrine action of the BM-hMSC secretome.
BM-hMSC regulate inflammation via IL-10 in the LTZ-induced PCOS mouse model
Our data showed that BM-hMSC engraftment reverses several key PCOS-related features such as insulin resistance, increased expression of androgen synthesis genes, a pro-inflammatory milieu, and abrogated fat metabolism. Notably, insulin resistance, androgen synthesis, and fat metabolism are all correlated with inflammation (57). We explored whether the effects of BM-hMSC treatment are mediated by anti-inflammatory factors within its secretome, such as IL-10. We first analyzed ovarian Il10 gene expression in all experimental groups. Interestingly, Il10 gene expression in ovary tissue was significantly higher in BM-hMSC-treated PCOS ovaries (5.37 ± 2.72 fold) compared with untreated PCOS ovaries (1.19 ± 0.46 fold; Fig. 6a). Moreover, IL-10 receptor gene (Il10r) expression in ovary tissue was also significantly higher in the BM-hMSC-treated PCOS group (2.13 ± 0.57 fold) compared with the untreated PCOS group (0.65 ± 0.17 fold; Fig. 6b). Several reports have demonstrated an increased pro-inflammatory milieu in fat tissues of PCOS women (58) and animal models (59). We assessed the impact of intra-ovarian delivery of BM-hMSC on white gonadal fat tissue inflammatory markers using qPCR. BM-hMSC treatment significantly downregulated Il6 (0.39 ± 0.04 fold), Il1b (1.0 ± 0.41 fold), Ccl2 (0.43 ± 0.02 fold), and Cd11c (0.45 ± 0.05 fold) expression in the fat tissue of BM-hMSC-treated PCOS mice versus that of untreated PCOS mice (Fig. 6g-j). Several inflammatory regulators, such as IL-10, IFN-γ, and TIMP-2, have been found to be lower in PCOS patients compared with healthy women (42, 60-62). We tested mouse serum using an antibody-based membrane assay and found that these cytokines were significantly lower in the untreated PCOS group (IL-10: 0.50 ± 0.08 fold, INF-γ: 0.79 ± 0.05 fold, TIMP-2: 0.82 ± 0.13 fold) compared with control mice (Fig. 6c). Importantly, these cytokines were significantly increased in PCOS mice after BM-hMSC treatment (IL-10: 1.20 ± 0.09 fold, INF-γ: 1.27 ± 0.11 fold, TIMP-2: 1.62 ± 0.22 fold; Fig. 6d-f). These results suggest that intra-ovarian injection of BM-hMSC has a systemic anti-inflammatory effect in the PCOS mouse model, likely mediated by IL-10 secretion from these cells.
Taken together, our data suggests that intra-ovarian injection of BM-hMSC reduces inflammation by increasing the expression of anti-inflammatory mediators such as IL-10 and its receptor in the ovary, and circulating IL-10, IFN-γ, and TIMP-2 in serum, while decreasing pro-inflammatory mediators such as IL-6, IL-1β, CCL2, and CD11c gene expression in periovarian adipose tissue.