Background
Polycystic ovarian syndrome (PCOS) is a complex, multifactorial endocrinopathy which affects approximately 4 to 10% of reproductive-aged women. Because it is a highly heterogeneous syndrome with a variable clinical presentation, criteria for diagnosis have been debated. Many authorities utilize the guidelines of Rotterdam/ASRM-sponsored PCOS Consensus Workshop Group [
1] and require the presence of at least two of the following: oligoovulation and/or anovulation, evidence of clinical or biochemical hyperandrogenism, and the presence of polycystic ovarian morphology during ultrasound examination.
PCOS is associated with several significant morbidities including infertility, obesity, insulin resistance, type 2 diabetes, dyslipidemia, and endometrial hyperplasia [
2‐
6]. Proposed etiologies for PCOS include hypothalamic-pituitary dysynchrony, aberrant gonadotropin pulsatile secretion, granulosa/theca cell dysfunction, and various metabolic derangements including exaggerated ovarian androgen production, hyperinsulinemia, and insulin resistance [
7‐
12]. Still, it is unclear whether the primary source of metabolic derangement is ovarian, hypothalamic/pituitary, or a combination of several systemic factors.
Several therapeutic options have been utilized to treat PCOS associated ovulatory dysfunction and infertility. These include weight loss, clomiphene citrate, exogenous gonadotropins, insulin sensitizers, and ovarian diathermy. Since its introduction as a treatment for type 2 diabetes in the United States in 1996, metformin also emerged as a common treatment for infertile women with PCOS [
13‐
15]. Despite widespread and continued use, the efficacy of metformin as a treatment for PCOS remains unproven and controversial. Metformin has been shown by some investigators to result in weight loss, normalization of menstrual cycles, and an improvement of conception rates following therapies such as ovulation induction and controlled ovarian hyperstimulation prior to
in vitro fertilization (IVF) [
16‐
19]. In contrast, other studies have demonstrated that metformin does not offer any clinical benefit [
20‐
22].
Metformin has been primarily characterized as an activator of AMP activated kinase (AMPK) [
23]. AMPK serves as a sensor of energy status at the cellular level and is activated by an elevated AMP/ATP ratio. Activation of AMPK may induce catabolic processes which generate ATP and reduce anabolic processes which consume ATP. It can also serve as an energy sensor in several organs. For example, small decreases in glucose result in AMPK activation and decreased pancreatic insulin production with increase hypothalamic-driven feeding behavior [
12,
24‐
27]. Moreover, AMPK has evolved in higher organisms to be a highly complex regulator of cytokine function where leptin and adiponectin activate AMPK in muscle to increase glucose uptake and fatty acid oxidation [
28,
29]. The significance of metformin's role as an AMPK modulator is uncertain in reproductive processes such as oocyte maturation, ovulation, and luteinization.
To date, there is limited evidence demonstrating a consistent physiologic effect of metformin on oocyte development, ovulatory function, and fecundity in animal models. Previous data in the bovine model have demonstrated that metformin results in inhibition of maturation of denuded (DO) and non denuded oocytes. A similar effect was seen with a specific AMPK activator (AICAR), implying that metformin's inhibitory action may be mediated in part by AMPK activation in the oocyte [
30]. Similarly,
in vitro studies using porcine oocytes have shown that metformin prevents the maturation of the oocyte when it is part of the cumulus oophorus complex (COC). However, it did not prevent maturation of the porcine DO [
31]. The AMPK activator, AICAR, has been shown to induce meiotic resumption in both mouse DO and COC
in vitro, whereas this effect is blocked by Compound C, a specific AMPK inhibitor [
32]. Metformin has also been shown to inhibit progesterone production
in vitro through an AMPK mediated pathway in a number of cell types derived from several different species [
33‐
35]. Notably,
in vitro metformin concentrations of all aforementioned studies were supraphysiologic (0.1 - 2 mM). According to Lee and Kwon [
36], serum concentrations in physiologic doses in humans are much lower, at approximately 8 - 10 μM.
Given the inconsistent results of published bovine and murine studies, and the controversy surrounding metformin's efficacy in PCOS-related ovulatory dysfunction and infertility, the goal of this study was to gain better insight into the effects of metformin on oocyte development and ovulation in mouse models which demonstrate metabolic and reproductive characteristics of women with PCOS. We utilized two different leptin mutant mouse strains. Both models, B6.Cg-m+/+
Lep
ob
/J or
ob/ob and the B6.V-
Lep
db
/J or
db/db), exhibit obesity, hyperphagia, a diabetes-like syndrome of hyperglycemia, glucose intolerance, elevated plasma insulin, and subfertility [
37,
38]. The
ob/ob strain does not produce endogenous leptin while the other strain,
db/db, possess a nonfunctional leptin receptor and has elevated systemic leptin levels. We hypothesized that metformin therapy will have an effect on oocyte maturation and/or ovulatory function in
ob/ob and
db/db animals compared to wild type (WT) mouse strains.
Discussion
Distinct features in women with PCOS, insulin resistance and compensatory hyperinsulinemia, lead to hyperandrogenemia due to increased ovarian androgen production and decreased production of sex hormone binding globulin [
42,
43]. Since hyperinsulinemia has been implicated as a significant cause of anovulation, many investigators hypothesized that a reduction of systemic insulin serum levels would result in an improvement of ovulatory function and overall fecundity in PCOS women. Initial studies investigating the use of metformin in PCOS demonstrated a beneficial role of metformin as an ovulation induction agent compared to placebo, clompihene citrate (CC), and CC and metformin combined [
16]. However, two subsequent large, prospective, double blind studies did not demonstrate any benefit for metformin treatment in women with PCOS in terms of ovulation rate and pregnancy outcome [
44,
45]. Despite a long track record of metformin use in type 2 diabetes, it still remains unclear whether it provides a beneficial reproductive effect as an adjuvant therapy in women with PCOS. Furthermore, if there is a beneficial reproductive effect of metformin, it is unclear whether it acts locally at the level of the ovary, pituitary, hypothalamus, or on a more systemic level. In this study, we have demonstrated for the first time, that metformin confers significant
in vitro and
in vivo effects on oocyte maturation in mouse strains with metabolic and reproductive characteristics of PCOS. Specifically, we demonstrate a reduction in the completion of meiosis 1 by oocytes
in vitro following metformin exposure in WT and
ob/ob mice, and an increase in the yield of mature oocytes and total overall oocytes following continuous dietary metformin for 7 weeks prior to superovulation in a
db/db in vivo model.
We hypothesized that treatment with the insulin sensitizer, metformin, would have an impact on oocyte maturation and/or ovulation in a PCOS-like mouse strain with a hyperinsulinemic and anovulatory phenotype. Previous studies examining the effects of metformin, have focused on specific compartments of the ovary, namely the oocyte and granulosa cells in WT animals (congenic mice and outbred strains of cows and pigs), with normal ovulatory function [
30‐
32].
In vitro studies have demonstrated direct effects of metformin on the ovary, which involve inhibition of basal and insulin stimulated granulosa cell P450 aromatase via MEK/ERK (MAPK kinase) activation [
46]. Similar to previously published studies detailing an inhibitory effect of metformin on
in vitro oocyte maturation [
30,
31], the results of this study demonstrated that metformin reduced
in vitro maturation of the mouse oocyte. Specifically, meformin exerted a significant reduction of maturation of oocytes derived from WT and
ob/ob mice, but not in
db/db mice. Notably, the
in vitro concentration of metformin which demonstrated this finding was at the highest concentration, and may represent an extremely elevated
in vivo serum level which surpasses the typical human metformin dose of 2000 mg daily dose (approximately 10 μM). These collective findings raise the possibility that this effect may be an artifact of toxicity of the high levels of metformin. Alternatively, these findings may be the result of
in vitro conditions, which may not be directly applicable to
in vivo conditions.
Based upon previous data [
47] which demonstrated an antiapoptotic effect of metformin on luteinized granulosa cells in PCOS patients undergoing IVF, it may be expected that metformin treatment would result in elevated progesterone levels in conditioned media from cultured granulosa cells derived from both transgenic mouse models which share PCOS characteristics. However, there was no obvious effect of increasing doses of metformin on progesterone levels in conditioned media derived from granulosa cells in any genotype. Therefore, it can be inferred that there was no significant change in cell number. The differences in our results may be attributed to species to species variability in response to metformin or may reflect the complexity of steroidogenesis, which likely involves multiple pathways independent of those regulated by metformin.
In vivo studies examined the chronic effects of metformin pretreatment on oocyte development and ovulatory performance in WT,
ob/ob and
db/db mouse strains during superovulation. With the use of 0.1 mg/ml metformin in drinking water (approximate to human dose of 2000 mg per day), these experiments demonstrated that metformin significantly increased the number of mature oocytes ovulated by 1.77 fold (p = 0.018) and the total overall number of oocytes released by 1.51 fold (p = 0.04) in
db/db mice during superovulation. Interestingly, this same result was not observed in the
ob/ob mouse strain, which shares many phenotypic similarities (obesity, hyperglycemia, hyperinsulinemia, and infertility with anovulation). In contrast to the
ob/ob mouse, which lack endogenous leptin production, the
db/db mouse has elevated systemic leptin levels. An explanation of the results seen only in the
db/db strain may be due to a possible effect of metformin on this animal's endogenously elevated leptin levels. Notably, there are preliminary data describing the reduction of leptin by metformin in women with PCOS [
48]. However, the fact that the
db/db mice lack a functional cognate receptor leptin receptor (long isoform) would imply that any change incurred by a decrease in leptin may be indicative of leptin eliciting a response through the less characterized short form of the OB receptor or via an unrecognized alternative receptor.
Given the known role of hyperinsulinemia and hyperandrogenemia in PCOS anovulation, it may also be initially inferred that the metformin treated
db/db genotype displayed improved glucose control and weight loss compared to other mouse strains. However, there were no significant differences in weight, glucose, or testosterone levels in any metformin treated mouse strain compared to controls. This observation in the
db/db mouse may signify a more pronounced, yet less detectable intrafollicular effect of hyperinsulinemia in this transgenic genotype. In line with prior observations of dysfunctional steroidogenesis and folliculogenesis in PCOS [
49], correction of this metabolic derangement with the insulin sensitizer, metformin, may have established a more favorable intrafollicular insulin environment and may have optimized ovulatory performance, resulting in an improvement in the production of mature oocytes during superovulation in the
db/db strain. Several authors have recently published findings which support a possible direct impact of metformin on the ovary. Stimulation of lactate production and activation of AMPK in granulosa cells by this compound has been proposed as a mechanism of improving follicular and oocyte development [
50]. Additionally, the findings of Palomba
et al. demonstrate a significant effect of metformin on intrafollicular insulin growth factor 2, several insulin growth factor binding proteins, estradiol, and androgen levels in women with PCOS [
51].
Although there is not a single ideal animal model for PCOS, several reproductive and metabolic features commonly observed in PCOS are present in the animal models utilized in the present study. As highlighted previously, there are other additional mouse and rat models which have been utilized to study PCOS [
52]. Unfortunately, some primarily possess metabolic traits, others demonstrate only reproductive characteristics, while others possess some combination of both [
37,
38,
52,
53]. As is true with other models, the mouse strains used in this study do not perfectly simulate human PCOS. To this end, one model will not be completely representative of all human PCOS phenotypes. Investigation in many different models will be likely required to gain a more comprehensive understanding of the metabolic and reproductive aspects of this syndrome. Since the
ob/ob and
db/db mice share both reproductive and metabolic characteristics of women with PCOS, it was most appropriate to utilize these strains to investigate the potential reproductive effects of metformin in a hyperinsulienmic and anovulatory
in vivo model. Although the exact mechanism of metformin has not been elucidated, it has been shown to be an activator of AMPK. The inhibitory effects of metformin at the level of the oocyte have been inferred from various mammalian studies using the AMPK activator (AICAR) and AMPK inhibitor Compound C [
30‐
32]. Unfortunately, it is difficult to directly assess the discreet physiologic role of metformin AMPK activation in reproduction in this model. In future studies, it may be possible to assess the role of the metformin AMPK pathway in another model since a group of investigators have demonstrated that the kinase LKB1 mediates glucose homeostasis in liver and the therapeutic effects of metformin [
54]. In order to definitively characterize the function of metformin via the AMPK pathway, the use of the LKB1 deficient mouse may provide additional insight into AMPK mediated local and systemic effects of metformin from a metabolic and reproductive standpoint.
Due to the wide variation of metabolic and reproductive characteristics in women with the polycystic ovarian syndrome, it has become a difficult task to identify if any PCOS phenotype may benefit from metformin. The unpredictable extent to which a specific end organ is affected by insulin resistance or hyperinsulemia (e.g. ovary of a woman with PCOS) is likely contributory to the inconsistent results of previous studies examining metformin use in PCOS [
54]. Given the continued uncertainty regarding the clinical reproductive benefit of metformin use for PCOS associated infertility, a study such as this, can assist the field in determining whether this adjuvant therapy is of tangible benefit in clinical practice. In the hyperinsulinemic and hyperandrogenic anovulatory leptin
ob/ob and
db/db mutant mouse strains, no significant effect of metformin was observed at physiologic levels
in vitro at the level of oocyte or granulosa cells to increase oocyte maturity or progesterone production respectively. As hypothesized, a beneficial
in vivo effect was demonstrated in the
db/db strain as seen by an improvement of the yield of mature oocytes during superovulation. When considering our findings, it may be reasonable to speculate that metformin may act to optimize oocyte development and production by the local and/or systemic reduction of hyperinsulinemia, androgen and leptin production, as well as by the reduction of inappropriately high intrafollicular estradiol levels (seen in PCOS patients) by attenuation of aromatase activity as highlighted previously [
46,
49]. In light of recent findings which suggest that metformin may act via an insulin dependent mechanism in the human ovary, this treatment may confer a significant effect on oocyte development and ovulatory performance in the
dbdb mouse and a subset of similarly hyperleptinemic and hyperinsulinemic women with PCOS [
55], Additionally, the larger follicular endowment of
db/db mice, compared to other genotypes, may also contribute an unknown influence on oocyte maturation and development during superovulation.
Authors' contributions
MES cared for all animals used in the study, performed a majority of all in vitro and in vivo experiments, and participated in manuscript preparation. All statistical analysis was performed by MES and reviewed by AKS and BRR. LG conducted in vitro progesterone assays and ovarian follicular counting experiments. MPL contributed to the conception and formulation of the study design and and to critical analysis of results. JOD assisted with animal care and progesterone assays. HJL served as an additional participant during the assessment of ovarian maturity in the in vitro studies, oocyte counts, and assessments during the superovulation studies. BRR participated in the conception and design of the study, critical analysis of data, and manuscript preparation. AKS is responsible for the original conception of the study, coordination and supervision of experiments, critical analysis of data, and preparation of the manuscript. All authors read and approved the final manuscript.